(Cambridge Studies in Comparative Politics) John M. Carey-Legislative Voting and Accountability-Cambridge University Press (2008)

This page intentionally left blank
Legislative Voting and Accountability
Legislatures are the core representative institutions in modern democracies.

Citizens want legislatures to be decisive, and they want accountability, but they
are frequently disillusioned with the representation legislators deliver. Political
parties can provide decisiveness in legislatures, and they may provide collective
accountability, but citizens and political reformers frequently demand another
type of accountability from legislators – at the individual level. Can legislatures
provide collective and individual accountability? This book considers what
both kinds of accountability require and offers the most extensive cross-
national analysis of legislative voting undertaken to date. It illustrates the
balance between individualistic and collective representation in democracies
and how party unity in legislative voting shapes that balance. In addition to
quantitative analysis of voting patterns, the book draws on field and archival
research to provide an extensive assessment of legislative transparency
throughout the Americas.
John M. Carey is the John Wentworth Professor in the Social Sciences at

Dartmouth College. He has taught at the Universidad Católica de Chile, the
University of Rochester, Washington University in St. Louis, Harvard Uni-
versity, and at the Fundación Juan March in Madrid, Spain. Carey’s most re-
cent books are Term Limits in the State Legislatures (2000, with Richard Niemi
and Lynda Powell), Executive Decree Authority (1998, with Matthew Shugart),
and Term Limits and Legislative Representation (1996). He has also published
articles in numerous scholarly journals as well as chapters in more than a dozen
edited volumes.
Cambridge Studies in Comparative Politics
General Editor
Margaret Levi University of Washington, Seattle
Assistant General Editor

Stephen Hanson University of Washington, Seattle
Associate Editors
Robert H. Bates Harvard University
Torben Iversen Harvard University
Stathis Kalyvas Yale University
Peter Lange Duke University
Helen Milner Princeton University
Frances Rosenbluth Yale University
Susan Stokes Yale University
Sidney Tarrow Cornell University
Kathleen Thelen Northwestern University
Erik Wibbels Duke University
Other Books in the Series

David Austen-Smith, Jeffry A. Frieden, Miriam A. Golden, Karl Ove
Moene, and Adam Przeworski, eds., Selected Works of Michael Waller-
stein: The Political Economy of Inequality, Unions, and Social Democracy
Lisa Baldez, Why Women Protest: Women’s Movements in Chile
Stefano Bartolini, The Political Mobilization of the European Left, 1860–1980:
The Class Cleavage
Robert Bates, When Things Fall Apart: State Failure in Late-Century Africa
Mark Beissinger, Nationalist Mobilization and the Collapse of the Soviet State
Nancy Bermeo, ed., Unemployment in the New Europe
Carles Boix, Democracy and Redistribution
Carles Boix, Political Parties, Growth, and Equality: Conservative and Social
Democratic Economic Strategies in the World Economy
Catherine Boone, Merchant Capital and the Roots of State Power in Senegal,
1930–1985
Continued after the Index

Legislative Voting and
Accountability
JOHN M. CAREY
Dartmouth College
CAMBRIDGE UNIVERSITY PRESS
Cambridge, New York, Melbourne, Madrid, Cape Town, Singapore, São Paulo
Cambridge University Press

The Edinburgh Building, Cambridge CB2 8RU, UK
Published in the United States of America by Cambridge University Press, New York
www.cambridge.org
Information on this title: www.cambridge.org/9780521884938
© John M. Carey 2009
This publication is in copyright. Subject to statutory exception and to the

provision of relevant collective licensing agreements, no reproduction of any part
may take place without the written permission of Cambridge University Press.
First published in print format 2009
ISBN-13 978-0-511-48077-5 eBook (NetLibrary)
ISBN-13 978-0-521-88493-8 hardback
ISBN-13 978-0-521-71191-3 paperback
Cambridge University Press has no responsibility for the persistence or accuracy

of urls for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
Contents
Preface page ix
1 TO WHOM ARE LEGISLATORS

ACCOUNTABLE? 1
1.1. Introduction 1
1.2. Decisiveness Problems 4
1.3. Collective versus Individual Accountability 7
1.4. Legislators, Principals, and the Structure of
Accountability 14
1.5. Plan of the Book 20
2 COLLECTIVE ACCOUNTABILITY AND
ITS DISCONTENTS 23
2.1. The Strong-Party Ideal 23
2.2. Legislative Parties and Discipline in Latin America 25
2.3. Trouble in Paradise: Partisan Representation
Falling Short 29
2.4. The View from the Chamber 36
2.5. The Shift toward Individual Accountability 40
3 THE SUPPLY OF VISIBLE VOTES 43
3.1. Visible Votes and Accountability 43
3.2. Who Can Monitor Votes? 49
3.3. The U.S. Experience 51
3.4. The Supply of Recorded Votes in Latin America 55
3.5. Conclusion 65
Chapter 3 Appendix 66
vii
Contents
4 DEMAND FOR VISIBLE VOTES 68

4.1.
Is Transparency Desirable? 68
4.2.
Incentives to Monitor and Publicize Votes 70
4.3.
How the Political Actors See Things 74
4.4.
Effects of Recorded Voting 83
4.5.
The Trend toward Visible Votes and Its Limits 90
5 COUNTING VOTES 92
5.1. Party Voting Unity and Collective Accountability 92
5.2. Measures of Voting Unity and Success 94
5.3. The Silence of Nonvotes 96
5.4. Data on Recorded Votes 102
5.5. Describing Voting Unity 107
Chapter 5 Appendixes 1–6 112
6 EXPLAINING VOTING UNITY 125
6.1. Legislative Parties and Institutional Context 125
6.2. Competing Principals and Existing Accounts of
Party Unity 126
6.3. Cohesiveness and Discipline: Weighted and
Unweighted Indices 128
6.4. Hypotheses: Legislative Parties and
Competing Principals 132
6.5. Picturing Party Unity across Systems 141
6.6. Models 146
6.7. Results 150
6.8. Extending the Analysis 159
6.9. Conclusion: Competing Principals Disrupt
Voting Unity 162
7 THE INDIVIDUAL-COLLECTIVE BALANCE 165
7.1. Transparency, Party Unity, Votes, and Accountability 165
7.2. Reviewing the Major Points 166
7.3. The Optimal Mix? 169
Appendix: Interview Subjects by Country 177

References 181
Index 195
viii
Preface
Shortly after the 2006 election, in which the Democrats recaptured control
of the U.S. Congress, the spoof newspaper The Onion ran a story in which
Nancy Pelosi, the new Speaker of the House, reprimanded her partisan
colleagues for supporting her legislative agenda without necessarily meaning
it. Referring to a fictitious bill, The Onion had Pelosi admonishing Democrats
not to ‘‘just pass it because I want it, but because you want it, too,’’ and went
on to describe Pelosi’s ‘‘concern that her relationship to the House was based
completely on voting’’ (The Onion, 42 [49], December 4, 2006).
Legislative decisions are about votes, and voting behavior is organized
by parties. If we want to understand legislatures and the representation they
provide, it makes sense to look at partisan voting. To The Onion, the joke
was that Pelosi might care about anything beyond that bottom line.
It never got big laughs, but I had a similar idea in mind around a decade
ago, when I started the project that became this book. At the time, the study
of voting in the U.S. Congress was a bustling cottage industry, but there
was almost no information about legislative voting outside the United
States. The reason, it seemed to me, had to be the lack of available data
on votes. So, to begin, I set out to collect data on votes in a number of
legislatures, mostly in Latin America where I had some experience, but also
in other assemblies where I could establish research connections. My first
surprise was that, in most countries, it was exceedingly unusual to record
how each legislator voted on a given proposal. What The Onion took to be
the bedrock of legislative representation could not be taken for granted in
many democracies.
As I explored the issue across more and more assemblies, it became clear
that a prior question – before how legislators vote – is whether assemblies
make it possible to know how legislators vote. So the research agenda
ix
Preface
evolved and expanded, and I spent as much time talking with politicians,
journalists, and activists about whether they favored voting transparency,
and why, as I did collecting and analyzing voting data.
As it turns out, I spent a lot of time on each, which accounts for the ten
years that passed between starting the project and publishing this book.
Those years have seen progress in the study of legislative voting beyond the
halls of the U.S. Congress. This book takes a step toward mapping, and
explaining, the world of partisan voting in legislatures. Data availability
remains an obstacle in most assemblies. Many still record few or no votes,
and those that do record often do not make vote records easy for outsiders
like scholars, or citizens, to examine. The problem is more than academic.
Lack of voting transparency is also an obstacle to accountability.
There is much more on this topic in the book itself. Here, I want to
recognize and thank the organizations and the people who made my re-
search possible. The book offers the broadest cross-national analysis of
recorded voting to date, and all the data collected for the project are avail-
able online for other researchers to use. Doing field research in ten coun-
tries, and collecting the data from fifteen others, required resources,
expertise, and effort beyond what I could muster on my own. Early
financial support was provided by National Science Foundation Grant
SES-9986219 and also by the Weidenbaum Center on the Economy,
Government, and Public Policy at Washington University in St. Louis.
I received outstanding research assistance at Washington University
from Christopher Kam, Connor Raso, Meg Rincker, John Bunyan, Sarit
Smila, Alba Ponce de León, Erica Townsend Bell, Gina Reinhardt
Yannitell, Adam Bookman, Rachel Kaul, Cheryl Boudreau, Juan Gabriel
Gómez Abellardo, Amy Nunn, and (now Senator) Jeff Smith. Jeff Staton’s
contributions are better described as collaboration than as research assis-
tance, and I continue to learn from Jeff. Rebecca Cantú provided solid
assistance during my brief visit at Harvard. At Dartmouth, assistance from
Anne Bellows, Justin Brownstone, Xavier Engle, and Seth Goldberg helped
bring the project across the finish line.
In the course of conducting field research and in collecting data from
assemblies far and wide, I drew on the expertise, and often on the hospitality,
of dozens of generous souls. Eduardo Alemán, Mark Jones, Valeria Palanza,
Roberto Sabá, and Mariano Tommasi shared data and provided insights into
Argentine politics. In Bolivia, thanks go to Diego Ayó, Carlos Cordero,
William Culver, René Mayorga, José Rivera Eterovic, and Eduardo Rodri-
guez. On Brazil, I am grateful to Barry Ames, Octavio Amorim Neto, Scott
x
Preface
Desposato, Argelina Figuereido, Wendy Hunter, Eduardo Leoni, Fernando

Limongi, Scott Mainwaring, Carlos Pereira, Timothy Power, David
Samuels, and Kurt Weyland. On Canada, New Zealand, and Australia,
Christopher Kam was, and is, the man. On Colombia, thanks to Luis
Fajardo, Ana Julia Ramos, and Elisabeth Ungar. In Costa Rica, Jorge Vargas
Cullel and Aixa Ansorena, Rafael Villegas Antillón, and Leo Nuñez Arias
competed for the title of most gracious hosts, and most insightful political
experts. In the Czech Republic, thanks to Elena Mielcova and to Daniel
Munich for sharing data. Felipe Cisneros and Andrés Mejı́a helped lead me
through Ecuador’s dense political thicket. In El Salvador, David Holiday
was another dual provider of safe haven and deep political knowledge. Eric
Voeten graciously shared voting data on France. On Guatemala, Harry
Brown Araúz, Javier Fortı́n, and Reginald Todd all offered key insights.
On Israel, the politics of which is pretty much self-explanatory to begin
with, Itai Sened and Doron Navot offered yet further clarity. Thanks to
William Heller for insights on the Italian parliament. In Mexico, Jeffrey
Weldon, Joy Langston, Cecilia Martı́nez-Gallardo, and Alejandro Poiré
all walked me through the politics of a democratizing legislature. In Nicar-
agua, Guillermo Garcı́a showed me the ins and outs of the Assembly and
provided critical contacts. In Peru, Cynthia Sanborn made me feel at home
and also led me through the shifting post-Fujimori political landscape. I was
happy to receive further help on Peru from Catherine Conaghan, Gregory
Schmidt, and Rick Walter. Steven Braeger, Sheila Espine-Villaluz, and Carl
Landé all shed light on the Philippines. In Poland, Wieslaw Dobrowolski
and Jacek Mercik shared data, and Meg Rincker provided on-the-ground
knowledge. On the Russian Duma, thanks to Moshe Haspel and Thomas
Remington. Thanks also to Manuel Alcántara in Salamanca, Spain, for mak-
ing data available from his Proyecto Élites Latinoamericanas. In Venezuela,
Ricardo Combellas explained the politics of Hugo Chavez’s (first) constitu-
tional overhaul, while Brian Crisp, José Molina, Steve Ellner, Janet Kelly,
Michael Coppedge, Miriam Kornblith, Berta Peña, and Juan Carlos Rey
delivered all-around expert observations.
Drafts of various papers that eventually formed parts of the book
benefited from critical feedback from participants in seminars at the Uni-
versidad de los Andes in Bogotá, the University of Chicago, Cornell, Duke,
Florida International University, the Fundación Juan March in Madrid,
George Washington University, Harvard, the Instituto Tecnológico
Autónomo de México, Notre Dame, Ohio State, Oxford, the Universidad
del Pacı́fico in Lima, Princeton, and the University of Vermont.
xi
Preface
Beyond these, a number of scholars provided comments and suggestions

on earlier papers or drafts that improved the final result. These include
James Alt, Joseph Bafumi, Gary Cox, Brian Crisp, Scott Desposato, Jorge
Dominguez, Kent Eaton, John Gerring, Jeanne Giraldo, William Heller,
Michael Herron, Simon Hix, John Huber, Jeffrey Jenkins, Mark Jones,
Nelson Kasfir, Jana Kunicova, Fabrice Lehoucq, Mona Lyne, Scott
Morgenstern, Kathleen O’Neill, Mark Payne, David Samuels, Matthew
Shugart, Peter Siavelis, Steven Swindle, Michael Ting, Jeffrey Weldon,
and Gerald Wright.
Thank also to Lew Bateman, my editor at Cambridge University Press;
Margaret Levi, the Comparative Politics Series editor; and to the anony-
mous reviewers they recruited. Lew and Margaret balanced consistent
support for the manuscript with sound critical judgment and equal meas-
ures of patience as I eventually made the necessary revisions.
After acknowledging intellectual debts far and wide, it is customary to
close a preface by paying homage closer to home. I turn to this task with
some apprehension, recognizing that the stakes are high. My office, after
all, is filled with books, and I have read all their prefaces, but I confess to
having studied the full contents of a much smaller number. So it stands to
reason that, for many readers, any lasting impression from Legislative Voting
and Accountability could depend on my eloquence regarding my family. I
admit straightaway I cannot do justice to that subject. My wife, Lisa, is the
greatest, and my sons, Joe and Sam, are too, for a million reasons that have
nothing to do with voting or accountability, although it is worth noting that
they indulge my habit of visiting legislatures in any state or country where
we travel, whether on vacation or for more important purposes, like soccer
tournaments. I really could not ask for anything more.
xii
1
To Whom Are Legislators Accountable?
1.1. Introduction
1.1.1. Overview
Legislatures are, formally, the principal policymaking institutions in mod-
ern democracies. The most fundamental policy decisions – budgets; trea-
ties and trade agreements; economic, environmental, and social regulation;
elaboration of individual and collective rights – all must be approved by
legislatures. What forces drive legislators’ decisions? What different polit-
ical actors place demands on legislators, and how do legislators’ actions
reflect these demands?
These are questions about what sort of representation citizens can expect
from those they send off to deliberate and make policy decisions on their
behalf. Citizens want legislatures to be decisive – that is, to resolve the issues
before them without chronic deadlock. They also want accountability, which
entails responsiveness on the part of legislators to citizens’ demands. In mod-
ern democratic legislatures, the principal vehicles for delivering decisiveness
are strong political parties. Decisiveness through party discipline, however,
presents a dilemma in terms of what kind of accountability is possible.
This book distinguishes between collective accountability and account-
ability that operates at the level of individual legislators, which often make
different demands on legislators. In modern democratic legislatures,
collective accountability operates primarily through parties and requires
legislators bearing a common party label to act in concert. Individual
accountability implies a more direct link between a legislator and citizens
and may require the legislator to act independently of party demands.
Individual accountability also requires that information about each
1
legislator’s actions be available to those outside the legislature and capable

of being monitored. Because the informational conditions for individual
accountability often are not met, maximum legislative individualism does
not necessarily produce individual accountability.
Scholarship on legislative accountability tends to regard collective ac-
countability favorably and legislative individualism with skepticism. Yet
surveys from legislators and citizens and the substance of political reforms
themselves in recent years all suggest that demand for individual account-
ability is strong, and technological advances have reduced the logistical
obstacles to making available the information necessary for individual
accountability. It is worth asking, then, whether individual accountability
is feasible and whether it can coexist with collective accountability, in what
measure, and under what conditions. One goal of this book is to examine
whether the vote records that legislatures produce are a critical ingredient
for individual accountability. A second goal is to use vote records to mea-
sure party unity, a key ingredient of collective accountability, and to explain
why some parties are more unified and others less so.
To be clear, transparency in the actions of individual legislators is a nec-
essary condition for individual accountability, and some measure of party
unity is necessary for collective accountability, but neither condition guar-
antees perfect accountability. Legislators whose every action is known may
still ignore their constituents’ demands, and unified parties may pursue
policies citizens abhor. This book merely suggests that individual account-
ability suffers when individual legislators’ actions are unknown to constit-
uents and that the failure of copartisans to act in concert undermines
collective accountability. Beyond these sorts of lapses, however, the tension
between individual-level and collective accountability is not fully reconcil-
able. Even full transparency offers access only to an accountability frontier
where an increase in individual accountability requires trading off some
measure of collective accountability, and vice versa. I reconsider this trade-
off in the concluding chapter, but the empirical substance of most of the
book examines whether the conditions exist to allow legislative represen-
tation to approach that frontier.
The book moves beyond previous research in the theoretical connection
it establishes between individual and collective accountability and in its
empirical scope. It illuminates the connection between legislative trans-
parency and accountability by examining why voting is transparent in some
legislatures but not others. It offers a simple and general account of the
various political actors institutionally empowered to place demands on
2
Introduction
legislators and how their relative influence affects legislative party unity.
This account, dubbed the competing principals model, generates hypoth-
eses tested against voting data from legislatures in nineteen countries. By
documenting what information is available about legislative votes and pro-
viding new tools to process the information, the book outlines the mix of
collective and individual accountability that legislators deliver across an array
of countries, as well as the potential for political reforms to alter that mix.
The rest of this chapter establishes vocabulary and concepts on which
the book depends. After defining some key terms used throughout the
book, I describe the unique role of political parties in organizing legislative
processes and as intermediaries of accountability between citizens and their
representatives. Then I contrast the ideals of collective and individual ac-
countability and discuss how electoral rules shape the balance between
collective and individual representation. Finally, I present the competing
principals model of demands on legislators and outline the plan of the
chapters that follow.
1.1.2. Definitions
1.1.2.1. Accountability. The expectation of accountability implies
a relationship between a legislator and some other actor or actors
(principals). Accountability means that legislators are responsive to the
preferences and demands of their principal(s), that information about
legislators’ actions is available to the principal(s), and that principals can
punish legislators for lack of responsiveness.
Accountability depends on professional ambition among legislators.
Professional ambition may be a purely venal desire for personal advance-
ment, or a purely altruistic desire to serve others by promoting policies that
advance some conception of the public interest, or some combination of
these. Whatever the motivation, ambition implies the desire to cultivate
electoral resources – renomination, or else nomination or appointment to
an even better office, campaign financing, and good favor among voters. It
also implies that legislators value access to resources within the legislature
itself, such as leadership positions, assignments to key committees, access
to support staff, big offices, perks, and such. Ambitious legislators curry
favor with political actors who can provide these key resources. The ability
to withdraw favor, and so deny the resources that fuel professional advance-
ment, is the enforcement mechanism behind accountability. Overall, ac-
countability should maximize legislative effort and responsiveness to the
3
principals’ preferences and minimize corruption and other abuses of power

at the principals’ expense.
1.1.2.2. Principals. Principals are political actors who command some

measure of loyalty from legislators, and whose interests a legislator might
represent and pursue in an official capacity. Given that most legislators in
democracies are popularly elected, we might think of voters as the ultimate,
universal principals to whom legislators are accountable. Under some
conditions, this is the case. Yet, as I argue in this book, political parties,
and specifically their leadership within legislative assemblies, are in many
cases the main principals that command legislator loyalty. In many
institutional settings, the level of accountability of legislators to voters
pales in comparison to their accountability to party leaders.
Beyond party leaders and voters, many political systems are populated
by other actors who, by virtue of their institutional positions, their orga-
nizational capacity, or other resources, can command the loyalty of legis-
lators. These include presidents, who are elected independently of the
legislature but are often endowed with resources and powers legislators
value or fear; governors in some federal systems, who may wield substantial
resources, including control over subnational political party machines;
interest groups, which direct electoral resources (funding, activist volun-
teers, mobilized voters); moneyed campaign contributors; and even those
in a position to bribe or extort politicians.
1.1.2.3. Decisiveness. Legislative decisiveness refers to the capacity of

legislatures to reach decisions on policy and to make those decisions
stick. Criticisms of legislatures frequently focus on failures along these
lines (American Political Science Association 1950; Sundquist 1981; Moe
and Caldwell 1994). Collective legislative accountability can be a solution
to legislative indecisiveness (Gerring, Thacker, and Moreno 2005; Gerring
and Thacker 2008). In this sense, decisiveness is central to the question of
whether legislative accountability is possible.
1.2. Decisiveness Problems
1.2.1. Bottlenecks and Cycling

Most national legislatures, and many subnational ones, are large assem-
blies, with diverse members numbering in the hundreds. Size and diversity
4
Decisiveness Problems
present a specific challenge. The number of policy options available in any

political environment is generally vast. Legislatures are supposed to pare
down the potentially infinite number of policy options available to a man-
ageable and coherent set of alternatives, among which a meaningful col-
lective decision can be reached. Failure to solve this decisiveness problem
may be the product of either too little legislative action or too much. In
either scenario, legislative scholarship envisions parties as a solution.
Cox (2006a) posits a ‘‘legislative state of nature’’ in which all members
have equal and unrestricted rights to speak on the floor on any issue, so
plenary time is unregulated. This state embodies a strong egalitarian norm
that privileges the ability of members to block assembly action over their
ability to trigger action, raising the specter of chronic legislative gridlock,
even in the face of pressing policy demands (Colomer 2001; Tsebelis 1999).
From this point of departure, Cox (2006a) notes that legislatures every-
where resolve the bottleneck problem with internal organization that
redistributes agenda powers unequally and that, in modern legislatures,
political parties consistently control access to the privileged agenda-setting
positions.
In the mirror image of Cox’s vision of unlimited filibuster and effective
unanimity rule, procedural rights in a legislature may be equally distrib-
uted, but rather than any legislator’s being able to block a vote on any
proposal, any proposal must be voted on. Now the decisiveness problem
becomes the potential for chronic instability of choice rather than the
inability to make any choice – that is, cycling occurs rather than bottle-
necks. The rationale here is well known. Formally, at least, most legislative
assemblies rely on simple majority rule for most decisions. Theoretical
characteristics of majority rule decision over multiple alternatives suggest
that failures of decisiveness could be characterized by general instability of
legislative decisions (Condorcet 1785; McKelvey 1976; Riker 1982). Yet,
even accounts of legislative politics that take the instability problem as
a point of departure frequently point to political parties as the key factors
that bring order to the potential chaos of majority rule (Laver and Shepsle
1996; Cox and McCubbins 1993).
In either the bottleneck-based or cycling-based account, parties are
credited with providing decisiveness by establishing privileged actors
who determine which proposals are debated and voted on and in which
order and, in doing so, make it possible for legislators to realize gains
unrealizable in unorganized, state-of-nature assemblies. The key point is
that, in almost all democratic systems, parties are the gatekeepers of the
5
formal offices that control action within the legislature. Moreover, Carroll,
Cox, and Pachon (2006) demonstrate that, as democracies mature, parties
expand their control over the offices that determine legislative activity, and
the distribution of these offices among parties grows increasingly regular.
In short, as party systems stabilize, so do the key partisan elements of
legislative organization.
1.2.2. Parties and Legislative Action

How does partisan control over the flow of legislative traffic provide deci-
siveness? Diverse accounts of legislative politics converge around the idea
that parties reduce the potentially infinite number of policy options to
a limited set, primarily by establishing platforms or manifestos that adver-
tise party positions to voters and then by disciplining legislators to con-
strain their voting in line with these party positions (Aldrich 1995).
Comparative studies of roll call voting suggest that legislative agendas
are strongly limited in ways consistent with the idea that parties produce
procedural order (Poole 1998; Cox, Masuyama, and McCubbins 2000;
Poole and Rosenthal 2001; Amorim Neto, Cox, and McCubbins 2003;
Rosenthal and Voeten 2004).
Because parties so consistently dominate legislative organization, it is
difficult to test the extent to which they account for the orderliness of
voting patterns. In a pair of ingenious studies, however, Jenkins (1999,
2000) compares voting in the Confederate Congress of 1861–65 with that
in the U.S. Congress during the same era. The legislatures were similar in
formal structure, in membership (many legislators served in both cham-
bers), and even in the issues on which they voted, but the Confederate
Congress was not organized along party lines. Voting patterns of Confed-
erate legislators were far less stable in important ways. First, voting coali-
tions were more fluid in the Confederate than the U.S. Congress (Jenkins
1999). Second, the ideological positions of Confederate legislators were
less stable over time (Jenkins 2000). Overall, the results suggest that polit-
ical parties impose order on voting in ways that make legislative decisions
predictable and stable.
Political parties may play this role in general, but even casual observers
will note that not all parties are equivalent. Comparative legislative scholar-
ship has long made much of the difference between strong and weak
political parties in controlling legislative outcomes. Yet, apart from abun-
dant analyses of the U.S. Congress, most of the legislative world has yet to
6
Collective versus Individual Accountability
be mapped in terms of party voting unity.1 If parties are highly unified in

their voting, then party labels can carry substantial policy content. That is,
citizens can observe just the partisanship of a legislator or a candidate and
infer what sort of policies she or he will support and oppose in office. If, by
contrast, parties are not unified, then this cue about political behavior
carries less information. Given that citizens tend to rely on low-cost cues
in evaluating politicians and parties, the reliability of party labels is a key
component of whether voters can be said to cast informed ballots in legis-
lative elections (Lupia and McCubbins 1998).
1.3. Collective versus Individual Accountability
1.3.1. Competing Visions

The discussion of decisiveness in legislatures implies the outline of a case
for collective legislative accountability through party-dominated represen-
tation. The key components of the case are as follows. Legislatures are
called upon to make decisions on a wide-ranging set of policies. The dif-
ficulties of collective decision making mean that no individual legislator can
credibly claim credit or responsibility for shaping policy on such a scale. In
contrast, political parties can both encompass a broad idea of the public
interest and plausibly claim to deliver policies that advance this idea. But
legislative parties can do this only if they are unified. Meaningful legislative
accountability, therefore, must be collective, through the organization of
legislatures by strong parties.
Yet, in the eyes of many political reformers, the idea of party-dominant
legislative representation has less appeal, and demands for accountability to
citizens at the level of individual legislator predominate. For example,
throughout Latin America in recent years a number of political reform
efforts have aimed to disconnect legislators from national party leadership
1
Levels and sources of party unity have been extensively examined in the U.S. Congress,
where scholarship has been preoccupied for more than a decade with parsing to what extent
levels of party voting are due to like-mindedness among copartisans (cohesiveness) versus
pressure from party leaders (discipline) (Krehbiel 1998; Cox and Poole 2004). For ‘‘un-
mapped’’ legislatures, the basic question of how unified parties are is of prior concern to the
cohesiveness-versus-discipline matter. Whatever its cause, some measure of voting unity is
necessary for party labels to convey information that is useful to voters. I take up the matter
of overall level of voting unity in Chapter 5 and the issue of cohesiveness versus discipline in
Chapter 6.
7
when the demands of leaders conflict with responsiveness to local constit-

uencies. Reform advocates describe popular disenchantment with disci-
plined parties directed by leaders who are insulated from punishment by
voters. In many cases, moreover, both the strong discipline and the insu-
lation of the leaders can be traced to a common source: electoral systems in
which control over candidate nominations is centralized among party lead-
ers and voters are allowed no means to distinguish among candidates within
a given party. The problem is most severe in systems of proportional rep-
resentation, where multiple seats are awarded in each electoral district.
The accountability dilemma here can be described as follows. As a pol-
itician advances within the party leadership, her access to power and perks
increases dramatically, but her electoral vulnerability decreases in a corre-
sponding manner because leaders occupy the top positions on closed party
electoral lists. This mitigates the leadership’s susceptibility to electoral
punishment, even if its party as a whole loses electoral ground. As a result,
the leaders who stand to gain the most from violating public trust and
pillaging state resources stand to suffer the least electoral indignity if their
party, collectively, is punished by voters. Rank-and-file politicians, whose
heads are the first to roll in any partisan electoral setback, might object
to being relegated to the marginal list positions that buffer their leaders,
but would-be rebels face a serious collective action problem in revolting
against their party leaders, because troublemakers can simply be removed
from the lists or demoted to perilous or even hopeless list positions by
the leadership.
The individualist dissent from the strong-party ideal implies a case for
accountability at the level of each legislator. The core of the argument rests
in the critique of party-dominated representation as imbuing the most
powerful legislative leaders with a sense of distance from voters that insu-
lates them from public disapproval. Instead, the argument goes, legislators
are most responsive to citizen demands when each is responsible for
cultivating her or his own support constituency, which in turn can reward
and punish its representative directly at the polls. Whereas advocates of
collective, partisan representation are primarily concerned with the ideo-
logical and policy content of party labels, the decisiveness of legislatures,
and the voters’ assessments of overall government performance (Powell
and Vanberg 2000), advocates of individual-level accountability are more
concerned with maximizing virtue – deterring the betrayal of the demands
of particular voters who picked an individual legislator as their represen-
tative (Persson, Tabellini, and Trebbi 2003).
8
1.3.2. Excursus on Electoral Rules: Iraq versus Afghanistan

The individualist dissent over legislative accountability focuses heavily on
the electoral link between citizens and their representatives – specifically,
on whether elections that foster collective representation can, or should,
be modified to foster individual accountability. Scholars, policymakers,
and journalists sometimes mistakenly equate the trade-off between
collective and individual representation with the distinction between
proportional representation and single-member district systems. Because
proposals for institutional design and reform often focus on elections,
it is worth considering how electoral rules do, and do not, map onto
the matter of collective versus individual representation. A comparison
of two contemporaneous, high-stakes cases of institutional design is
useful for illustration.
Amid all the debates surrounding the regime changes in Afghanistan and
Iraq during the middle years of this decade, one of the less voluble, but
nonetheless crucial, was the discussion among both policymakers and aca-
demics over how to craft mechanisms to represent diversity in each coun-
try’s new legislative assembly.2 U.S.-led invasions of Afghanistan in 2001
and Iraq in 2003 had produced governments commissioned to craft new
constitutions and to hold elections to fill the political offices so founded. In
both cases, there was widespread acknowledgment that plural societies war-
ranted representation of broad diversity within the legislature. A fundamen-
tal challenge in both cases was to identify what sort of diversity ought to be
privileged in legislative representation. Various dimensions of representa-
tion, including geography, ethnicity, religion, and gender, were promi-
nently on the table in each case. Less widely noted was that the debates
over how best to move toward electoral democracy in Afghanistan and Iraq
embodied the fundamental trade-off between collective and individualistic
representation in contexts relatively unbound by existing precedent.
1.3.2.1. Iraq. Iraq elected two legislatures in 2005 – first, in January,

a transitional dual-purpose parliament and constituent assembly, the
main task of which was to draft a constitution to be ratified in an
2
The brief discussion that follows here of Iraq and Afghanistan is not meant to serve as
a thorough review of legislative electoral rules, much less as a comprehensive analysis of
the politics of these countries. The former is provided in an impressive literature on com-
parative electoral systems (Duverger 1954; Taagepera and Shugart 1989; Lijphart 1994;
Cox 1997), and the latter is well beyond my capacity.
9
October plebiscite; then a National Assembly to serve a full four-year term

under the new constitution. Notwithstanding some subtle but important
modifications to the electoral rules between the two elections, the central
characteristic of both Iraqi elections was strong collective representation.
The electoral law for January, crafted primarily by the United Nations
Electoral Assistance Division and handed down as law by the outgoing,
U.S.-led Coalition Provisional Authority, stipulated that the entire country
encompassed a single electoral district with 275 seats, the implications of
which were far-reaching for the types of legislative representation possible
in Iraq (Dawisha and Diamond 2006).3
First, the high district magnitude effectively mandated that elections
would be based on closed lists. Voters would not have the option of
casting preference votes for individual candidates. Second, high magni-
tude made it possible to award legislative seats to lists that won relatively
small vote shares overall, thus allowing for a high degree of proportion-
ality. Third, a nationwide list system like Iraq’s does not favor any
predetermined concept of representation – for example, geographical,
ethnic, or religious – but rather simply rewards lists that can mobilize
the most voters. However, because the composition of the assembly is
determined as much by the selection of candidates as by the popular vote,
such a system also opens up the possibility of adopting measures that
might tip legislative representation toward categories of candidates dis-
advantaged in a more individualistic electoral marketplace. Specifically,
in the Iraqi case, gender quotas for candidates mandated that every third
candidate must be a woman.
Both of the Iraqi elections in 2005 produced assemblies in which
twelve separate lists won representation. The effective number of seat-
winning parties was 3.14 in January and 3.45 in December (Laakso and
Taagepera 1979), and the elections were marked by a close correspon-
dence between votes cast and seats awarded to each list, with a Gallagher
Disproportionality Index of less than 3 percent both times (Gallagher
1991) and substantial representation of ethnic groups previously margin-
alized in Iraqi politics (Burns and Ives 2005). The guaranteed placement
of women at regular intervals on closed lists translated into assemblies
with 29 and 26 percent of women overall – almost twice the worldwide
3
One compelling motivation for this choice had to do simply with logistics of electoral
administration: Iraq lacked a reliable census by which legislative seats might be apportioned
across districts according to population.
10
average (Inter-Parliamentary Union 2006). In sum, by the benchmarks

widely used in legislative studies at least, the Iraqi system realized many of
the normative goals associated with the representation of diversity at the
collective level.
1.3.2.2. Afghanistan. The Afghan experience with establishing

a national assembly was substantially different. An indirectly elected
assembly drafted a new Afghan Constitution that was ratified in early
2004 and stipulated the popular election of both a presidency and
a bicameral legislature later that year. The presidential election took
place on close to the original schedule, in October 2004, but legislative
elections were twice postponed, in part because of the logistical challenges
of conducting elections that simultaneously honor the determination of the
Afghan government to:
guarantee an element of regional representation via geographical
districts;
avoid a winner-take-all system of elections in which only the top party
or candidate in a district wins representation;
ensure voter choice over individual legislative candidates; and
guarantee the representation of women (Johnson, Maley, Thier, and
Wardak 2003).
Although critics sustained objections, and uncertainty about the
electoral system remained into the early months of 2005, Afghan president
Hamid Karzai ultimately settled on the single nontransferable vote
(SNTV) for the September election (Reynolds 2006). SNTV, currently
used only in Taiwan, Jordan, and Vanuatu, but most familiar for its long
use in Japanese elections, from 1958 to 1994, is plurality rule in multimem-
ber districts. Each voter casts a ballot for her or his first-choice candidate,
and the candidates with the most votes are elected in each district, up to the
number of seats available. SNTV is attractive in its simplicity, and for its
potential to allow minority groups to secure representation while simulta-
neously holding out the promise of a bond of direct personal accountability
between voters and their representatives.
SNTV, however, is subject to at least two severe drawbacks that
undermine its potential to provide viable representation in the Afghan
context. First, SNTV presents any collective political actor – a party, for
example – with a formidable coordination problem in translating electoral
support into legislative representation. The problem is a fundamental
11
conflict of interests between the party and its individual politicians.4 Parties
seek to win as many seats as possible. Individual politicians may prefer to be
members of strong parties, but their first priority is to win office. Under
SNTV, candidates who seek to minimize the risk of individual defeat have
incentives to draw votes away from copartisans, undermining the collective
goal of translating votes to legislative representation efficiently. By
privileging electoral individualism, SNTV presents formidable challenges
to parties’ ability to foster internal cooperation among politicians and thus
to provide collective representation (McCubbins and Rosenbluth 1995;
Cox and Thies 1998; Marlowe 2007).
An even more immediate challenge to the feasibility of SNTV in Afgha-
nistan was the incompatibility between individualistic legislative represen-
tation and the representation of women. The Afghan Constitution requires
that at least two lower-house legislators from each of the country’s thirty-
four provinces be female (Art. 83). SNTV provides no alternative basis
other than individual vote totals for awarding legislative seats, so unless
at least two of the top candidates in each province were women, the Afghan
system requires bypassing male candidates with more votes in order to seat
female candidates with fewer votes. In a society where gender-based
inequalities in personal resources, as well as gender bias among voters,
constrain the viability of female candidates, this outcome was inevitable.
In the September 2005 election, 19 women were elected on the basis of
their vote totals alone, but 49 additional women were awarded seats in the
Loya Jirga despite having won fewer votes than 422 other male candidates
(Reynolds 2006). In sharp contrast to Iraqi gender quotas, the purely per-
sonalized and individualized character of legislative voting in Afghanistan
throws into stark relief the mechanism by which quotas delivered these
women to their seats while male candidates with higher vote totals lost.
The fundamental contrast in the Iraqi and Afghan choices over electoral
rules, at this point, is between privileging collective and individualistic
representation. For myriad reasons, the system chosen in Iraq leans toward
the former. This facilitated the initial, descriptive representation of various
collective identities – most notably by party alliance, ethnicity, religion,
4
The problem is also increasingly severe as district magnitude increases. Magnitudes in
Japanese SNTV elections ranged from three to five. In Afghanistan, the median district
magnitude for parliamentary elections was seven, with a third of districts electing ten
or more representatives, and the largest, Kabul, electing thirty-three (Constitution of
Afghanistan, Article 82; Reynolds 2006).
12
and gender. Afghan rules lean toward privileging connections between

voters and individual candidates. Trying simultaneously to guarantee rep-
resentation according to at least one prominent form of collective identity,
gender, presents a formidable challenge.
1.3.3. Collectivism versus Individualism, Not Proportional

Representation versus Single-Member Districts
The Iraqi and Afghan cases suggest that the distinction between individu-
alistic and collective representation can be more important than the
principal characteristic by which electoral systems are more frequently
distinguished – whether elections are winner-take-all in single-member
districts (SMD) or are based on proportional representation (PR). The
characteristics and relative merits of SMD versus PR are central to
a long-standing literature on legislative elections, whose predominant con-
clusion has been that PR elections are normatively superior to SMD elec-
tions (Sartori 1976; Lijphart 1994; Huber and Powell 1994; Colomer
2001). This conclusion rests on two key assumptions, however: that polit-
ical parties are fundamental units of legislative representation and that
a left-right spectrum meaningfully describes the ideological arena of party
competition. In the long-standing industrialized democracies, where most
studies of legislative representation have been conducted, there is solid
empirical evidence for these assumptions (Powell and Vanberg 2000). They
are open to greater skepticism in other environments, though, particularly
where party systems are more volatile or party reputations less stable.
The point here is that the foundation on which the conventional SMD
versus PR debate has been conducted is weak in many political environ-
ments where the most critical choices about how to organize legislative
representation remain open. The Iraqi and Afghan cases, in which estab-
lished party systems are completely absent, are extreme examples, but it is
worth noting that SMD versus PR was not central to the debate in either
context; winner-take-all rules gained traction in neither case (Dawisha and
Diamond 2006; Reynolds 2006). Rather, the critical distinction in these
cases is over whether electoral rules ought to prioritize collective versus
individualistic representation. This theme has been central to debates over
reforming legislative representation much more widely during recent dec-
ades, particularly with respect to mixed-member electoral systems that
combine SMD with list PR elections within the same legislative chamber,
variants of which were adopted in the 1990s by more than twenty countries
13
(Shugart and Wattenberg 2001; International Institute for Democracy and

Electoral Assistance 1997; Culver and Ferrufino 2000).
Legislatures offer the promise of representing the diversity of the polity,
but electoral rules affect the dimensions along which diversity can be trans-
lated into representation. Although the differences between SMD and PR
elections have traditionally been essential to the study of comparative
legislatures, this distinction is growing less central relative to that between
individualistic and collective representation, which is quite a different mat-
ter, both theoretically and empirically (Carey and Shugart 1995). Whereas
the literature on comparative legislative representation tends to favor
PR over SMD, there is less academic consensus on the relative merits of
individualistic versus collective representation (Golden and Chang 2001;
Persson and Tabellini 2003).
1.4. Legislators, Principals, and the Structure of Accountability
1.4.1. Party Leaders and Everyone Else

At the beginning of this chapter, I defined accountability in terms of the
relationship between a legislator and a principal or principals who control
resources the legislator values and so command loyalty. The discussion of
decisiveness that followed illustrates why parties are ubiquitous in legisla-
tures and why they control important resources. Thus, legislators always
and everywhere confront, in the leadership of their own party, a key prin-
cipal in a position to impose demands on their behavior. For legislators in
many environments, moreover, party leadership is pretty much the only
principal that matters. In electoral systems where voters cast ballots only
for the party, and where party leaders control access to their own lists, there
is no direct link between legislators and voters. The Iraqi case is an exam-
ple, but it is far from unique (Lijphart 1999).
In contrast, for a legislator elected by a purely personal vote, the support
constituency is clearly a primary principal, but even such a legislator gen-
erally confronts two principals – voters and party leaders – because party
leaders control resources within the legislature itself even when electoral
rules encourage individualism. The extent to which legislative individual-
ism predominates over collective, partisan representation depends on the
relative value of the resources that are controlled by either the voters or the
party leaders, and the propensity of voters and party leaders to want dif-
ferent things and thus pull ‘‘their’’ legislators in opposite directions.
14
Legislators, Principals, and the Structure of Accountability
In short, because of the ubiquity of parties, even the individualistic

vision of legislative representation tends to involve (potentially) compet-
ing principals: voters who pick an individual legislator, and the party (or
bloc, or group, or coalition, etc.) with which the legislator aligns in the
assembly. Yet the potential for principals to compete for legislator loyalty
and the effects of this on legislative individualism go beyond party versus
electoral constituency. In many political systems, other actors control at
least some resources that affect the legislative process or the ambitions of
legislators.
Most prominent here are presidents, who are elected independently
from legislators but who are often constitutionally endowed with legisla-
tive authorities, such as vetoes, decree- and rule-making powers, or the
ability to offer or amend legislative proposals under restrictive rules.
Presidents also control appointments to public offices and may have dis-
cretion over the release of budgetary funds for public projects. In many
democracies, the list of resources controlled by presidents and valued by
legislators is extensive. As a result, the possibility for presidents to exert
pressure on legislators to act in ways contrary to party, or constituent,
demands is substantial.
Though to a lesser extent than presidents, governors in some federal
systems control significant resources valued by national legislators. In
Argentina, for example, provincial governors often control their parties’
nominations, including those for incumbent legislators who aspire to re-
election or to election to some other office. In Brazil, governors control
appointments to state-level cabinet posts that are widely sought by national
legislators. Like presidents, governors may exert pressure on legislators in
ways contrary to the demands of the party at the national level.
The list of other principals who potentially compete with party leaders
for the loyalty of legislators is not limited and could include interest groups,
political activists, those who fund campaigns, and those who use bribery or
extortion to induce compliance. My focus in this book is on the most
prominent and prevalent principals who exert pressure on legislators in
the widest range of contexts: party leaders, voters, and, in systems where
the chief executive is directly elected, presidents.
1.4.2. Competing Principals and Individualism

Let us combine the themes discussed in this chapter to consider the relation-
ship between the various principals to which legislators might be accountable
15
and the ideas of collective-versus-individualistic representation. Given the

prominence of strong parties as a normative ideal in much of the literature on
legislatures, and in light of the empirical fact that democratic legislatures are
organized along party lines, I rely on party unity as a touchstone, a funda-
mental metric. From that conceptual point of departure, I regard legislative
behavior in terms of its deviation from party unity.5
In terms of legislative principals, representation is party-dominant when
the party leadership is the only political actor to which legislators are di-
rectly accountable. This occurs when central party leaders control nomi-
nations for legislative office and list positions (if more than one legislator is
elected from a district). That is, voters are not afforded the opportunity to
select from among various legislative candidates within a party. Under
these conditions, party leaders control not only resources interior to the
legislature but also the key electoral resources on which a legislator’s
career depends. Voters can reward and punish parties collectively for their
positions and performance, and so can be regarded as principals of the
parties via elections, but they have no direct say over the fate of individual
legislative candidates. Under such conditions, accountability is entirely
collective, at the party level. Given the undiluted influence of party leaders
over legislators, party unity ought to be high under these conditions. The
relationships among legislators, parties, and voters are portrayed in Figure
1.1, where the arrows indicate control by a principal with the political
resources to reward and punish.
VOTERS
PARTY
LEGISLATOR
Figure 1.1 Party-dominant representation.
5
In Chapter 6, I also consider broader legislative coalitions – government versus opposition,
for example – but parties are the component units of such coalitions, so this is simply
a matter of moving to a higher level of aggregation.
16
Next, consider a political system where voters have the ability to reward
and punish individual legislators directly, perhaps because primary elections
determine nominations, or because party lists are open and candidates win
legislative seats according to their individual preference votes, or because
there are no party lists at all and multiple candidates from each party run
either in a free-for-all format or under a transferable vote rule.6 Under any
of these circumstances, the voters are a legislator’s direct principal. Because
party labels are generally attached to the candidates for whom voters vote,
and because in the aggregate a party’s fortunes depend on the success of its
candidates, voters are also indirectly principals to the parties. Meanwhile,
party leaders, in all likelihood, remain important principals for legislators,
to the extent that they control resources within the assembly itself that
legislators value. They may also retain control over electoral resources, such
as influence over nominations and financing. Thus, legislators now confront
two principals, who may well make competing demands (see Figure 1.2).
We could add a directly elected president to either of these scenarios, as
in Figure 1.3. The formal powers of presidents over the legislative process
vary enormously, but most control access to coveted appointed posts, and
many are endowed with authority to shape the legislative agenda directly,
to veto all or parts of bills, and to offer counterproposals to legislative
initiatives (Shugart and Carey 1992; Aleman and Tsebelis 2005). The array
of powers of most directly elected presidents provides substantial leverage
with which to influence legislative behavior. As with a direct electoral
VOTERS
PARTY
LEGISLATOR
Figure 1.2 Parties and voters as competing principals: Individual-predominant
representation.
6
See Carey and Shugart (1995) for more details on the variety of electoral systems and their
relationship to the relevance of the personal vote, as well as for empirical examples.
17
VOTERS
PARTY PRESIDENT
LEGISLATOR
VOTERS
PARTY PRESIDENT
LEGISLATOR
Figure 1.3 Presidents as competing principals with party leaders.
connection to voters, presidential influence adds another, potentially com-

peting, pressure to that exerted by party.
The list of potential principals placing demands on legislators could
expand, and some research on legislatures has explored specific examples.
Carey and Reinhardt (2004) examine the influence of state governors on
their national-level legislative copartisans in Brazil. Unlike presidents, the
governors do not exercise direct authority over the national legislative
agenda, but they do direct the flow of many resources that are essential
to legislative reelection prospects, and they control access to state-level
appointed posts that, in Brazil’s decentralized system, are attractive to
many national legislators. Hix (2002) demonstrates that members of the
European Union Parliament experience varying levels of competing pres-
sures from the leadership of their party blocs within the EU Parliament
itself and from the national-level parties from which they are elected in
18
Brazil
VOTERS
PARTY PRESIDENT GOVERNOR
LEGISLATOR
European Union Parliament
VOTERS
NATIONAL PARTY EU PARTY
EU LEGISLATOR
Figure 1.4 Increasingly complex sets of principals.
their countries of origin. Figure 1.4 suggests how such additional principals
might map onto the accountability relationships described so far.
As the number of principals grows, the potential for competing pres-
sures to pull legislators away from solidarity with their party increases, and
we should expect party unity to drop.
Does the lack of party unity indicate the existence of legislative in-
dividualism? I interpret it as such, for a couple of reasons. One is that
deviance from the party line often indicates an effort by a legislator to act
on behalf of a constituency of citizen supporters, independently from any
mediating forces. This story is associated with the idea of the personal
vote and of individual-level accountability in the simplest sense (Cain,
Ferejohn, and Fiorina 1981; Carey and Shugart 1995). But what if a leg-
islator is pulled away from the national party by another institutional
19
political actor – by the president, for example? Then deviance from party
loyalty might not be regarded as simply a matter of individual volition.
Yet, as Sophocles’ Antigone discovered, cross-pressures, by their nature,
turn the decisions individuals make into acts of self-definition. In this
light, a legislator who lines up with the president (or governor, or whom-
ever) in contradiction to her legislative party leadership is staking out
a position, even if reluctantly, for which she can be held individually
accountable. Finally, although the number of potential principals making
demands is not limited, legislative decisions almost always come down to
binary choices – voting aye or nay. We may sometimes know what side
a president is on, or a governor, or an interest group. Less reliably still can
we draw any inferences about what the legislator’s constituency support-
ers demand. We can know, however, what side a legislator’s party is on by
how her copartisans voted.
In short, party unity is the point of reference because legislatures every-
where are partisan, because legislators everywhere answer to party leaders
as principals, and because party unity can be identified by the fundamental
act of legislative decision making – that is, voting. Where party unity is
lacking, there are various stories we can tell about legislative motivation,
but they all involve legislators’ making decisions to deviate from the game
plan of the team – the collective unit – that is the central basis of legislative
organization. It is no accident that the term ‘‘party line’’ has entered our
vernacular to connote the antithesis of individualism and independent
thinking. I contrast party unity and collective representation, then, with
disunity, which I associate with legislative individualism. The type of ac-
countability that is possible in any given legislature depends on what sort of
representation is provided.
1.5. Plan of the Book

The next chapter discusses collective, party-dominated representation as
an ideal, as well as in practice in a range of (mostly) Latin American legis-
latures. I note the high regard for the idea of collective representation in
academic opinion but also the contemporary decline in confidence in po-
litical parties among the public. I review various reforms aimed at reducing
the influence of central party leaders in Latin America and survey evidence
from legislators in fifteen Latin American countries indicating their skep-
ticism about centralized authority in parties and their devotion to the ideal
of individualistic representation.
20
Plan of the Book
Chapters 3 and 4 explore the conditions under which accountability

at the level of individual legislator is possible, focusing on the level of
transparency in the fundamental decisive action in all legislatures: voting.
Chapter 3 reviews the history of individual accountability through voting
records in the United States and then examines the supply of recorded votes
throughout Latin America. Chapter 4 shifts the focus to the factors that
create, or resist, demand for transparency in legislative voting. Drawing on
interviews and primary and secondary documentary sources, I argue that
the demand for recorded voting comes most consistently from opposition
legislators, dissidents within governing parties and coalitions, and outside
groups with an interest in monitoring legislative action. The basic logic is
that legislators’ most ubiquitous principals, party leaders, have a structural
advantage – proximity – in monitoring legislative voting. If other potential
principals are to make effective demands on legislators, they need to be able
to monitor as well, and that requires externally visible votes. Legislators
who want to appeal to outside principals also have an interest in recording
votes. Despite resistance from those who control legislative agendas, pres-
sure for transparency from these sources, combined with technological
changes that reduce the costs of recording and publishing, is steadily in-
creasing the supply of recorded votes.
The empirical material in Chapters 2, 3, and 4 is drawn overwhelmingly
from Latin America for a couple of reasons. First, I am interested in the
prospects for individual accountability as a normative rival to collective
accountability; individualistic representation is greater in presidential
systems than in parliamentary systems; and the greatest concentration
of presidential regimes is in the Americas. The political systems of the
Americas represent ‘‘most likely cases,’’ in that if we are to observe indi-
vidualistic representation and individual-level accountability anywhere, it
is likely to be there. The presidential systems of the Americas may, there-
fore, exhibit a bias toward legislative individualism, but they should also
provide a good window on the variety of ways in which this sort of behavior
manifests itself – an attractive property for this mostly qualitative portion of
the book. The second reason is purely pragmatic – I am better able to
conduct field research in Latin America than in many other parts of the
world, because of language and access to sources.
Chapters 5 and 6 turn to the quantitative examination of recorded votes
from legislatures in nineteen countries. Here the empirical scope grows
broader, in that I draw on recorded vote data from systems beyond the
Americas, but also somewhat shallower, in that no recorded votes at all are
21
available for most legislatures, and the records that are available are gen-
erally light on context beyond the digital ones and zeros of aye and nay
votes. Chapter 5 describes the recorded vote data, as well as the various
indices I use to turn the vast matrices of legislators’ aye and nay votes into
statistics that describe either voting unity at the party or coalition level or
the alternative, legislative individualism. That chapter also presents statis-
tics that describe the levels of party unity in the various political systems for
which I have collected recorded votes.
Chapter 6 develops and estimates a statistical model of the factors that
affect the diversity of principals to which legislators are accountable. The
analysis shows that party unity is highest when party leaders are the dom-
inant principal – that is, in parliamentary systems where voters do not cast
personal preference votes among candidates. Factors that subject legisla-
tors to influences from additional principals can diminish voting unity.
Electoral rules that provide for a personal vote among copartisans have
this effect. The presence of an independently elected president can also
reduce unity, particularly among legislators allied within the president’s
own party and governing coalition. Presidents, in short, are as much lia-
bilities as assets to their legislative allies.
The final chapter concludes with observations about the distinction
between legislative individualism and individual accountability, and the
potential for compatibility between the latter and collective accountability.
22
2
Collective Accountability and Its Discontents
2.1. The Strong-Party Ideal

The normative desirability of strong-party government is often taken
as axiomatic among academics. In 1950 the American Political Science
Association published a widely read report urging reforms to strengthen
the two major U.S. parties in the name of enhancing collective account-
ability, or what the APSA called ‘‘responsible party government.’’ In
doing so, the APSA was itself hearkening back to a vision of party-led
parliamentary government espoused almost a century earlier by the British
journalist and scholar Walter Bagehot (1867). As the APSA (1950:1) put it,
‘‘An effective party system requires, first, that the parties are able to bring
forth programs to which they commit themselves and, second, that the
parties possess sufficient internal cohesion to carry out these programs.’’
The report, moreover, explicitly linked the party unity that makes
possible collective accountability to the ability of the national party organ-
izations to cultivate control over the sort of electoral resources that would
make them stronger principals to congressional candidates:
As for party cohesion in Congress, the parties have done little to build up the kind of
unity within the congressional party that is now so widely desired. Traditionally
congressional candidates are treated as if they were the orphans of the political
system, with no truly adequate party mechanism available for the conduct of their
campaigns. Enjoying remarkably little national or local party support, congressio-
nal candidates have mostly been left to cope with the political hazards of their
occupation on their own account. A basis for party cohesion in Congress will be
established as soon as the parties interest themselves sufficiently in their congres-
sional candidate to set up strong and active campaign organizations in the constit-
uencies. (1950:21–22)
23
Current observers might argue that, in the closing decades of the

twentieth century, both the Republican and Democratic parties in the
United States heeded the APSA’s advice in developing formidable candi-
date recruitment and fundraising organizations at the national level. In-
deed, voting unity in both parties rose substantially ( Jacobson 2000; Lowry
and Shipan 2002). These developments, of course, have been greeted with
criticism by many, and calls for reform, much as the looser, midcentury
congressional party system did. Yet recent criticism is generally aimed at
strategies of campaign finance, whereas the broader ideal of unified parties
and collective responsibility retains solid academic support (Corrado 2002;
Mann 2003; La Raja 2003).
The norm is even more widely held among academic observers of legis-
latures outside the United States, according to a recent study of discipline
throughout Europe:
The maintenance of a cohesive voting bloc inside a legislative body is a crucially
important feature of parliamentary life. Without the existence of a readily identifi-
able bloc of governing politicians, the accountability of the executive to both
legislature and voters falls flat. It can be seen, then, as a necessary condition for
the existence of responsible party government. (Bowler, Farrell, and Katz 1999:3)
Wrapping up a broad survey of the state of political parties in Latin

America in the 1990s, Mainwaring and Scully (1995:473–74) lament the
tendency of presidents to campaign and govern based on personalistic
appeals rather than by cultivating stable party support:
As electoral democracy becomes accepted as the mode of forming governments in
most Latin American countries, and as the enormous costs of weak party systems
become apparent, perhaps leaders will pay more attention to the challenge of build-
ing democratic institutions and will govern through parties and with them. Without
a reasonably institutionalized party system, the future of democracy is bleak.
In short, strong parties have long been held in high academic esteem. In
the next section, I describe how legislative parties in a variety of Latin
American legislatures where I have conducted research reach and enforce
collective decisions. This is a description of the mechanics by which the
collective vision of representation through parties might be realized. Po-
litical reformers often see things differently. It is not that they aspire to
feckless parties. Nor, indeed, would most academics who call for strong
parties aspire to Leninist centralism. But whereas academic observers have
been inclined, on the whole, to see parties as weaker than they ought to be
and needing fortification, the general tide of reform in Latin America has
24
Legislative Parties and Discipline in Latin America
run against the authority of central party leaders, in the name of increasing
the accountability of individual legislators.
2.2. Legislative Parties and Discipline in Latin America
2.2.1. How Parties’ Positions Are Determined

All legislatures with which I am familiar are organized along party lines,
meaning that party units are accorded rights over legislative resources,
including representation on the organ that controls the legislative agenda,
as well as whatever offices and staff are available. Party groups in Latin
American legislatures are variously known as fracciones, bancadas, or grupos.1
The norm among legislative party groups in Latin America is to meet
at least weekly when the legislature is in session to discuss the upcoming
agenda and to establish both whether there is to be a group position on
each issue and what those positions will be. Party groups are subordinate
to national party organizations and generally can be instructed by them as
to how to vote on specific issues.2 National party congresses invariably
occur less frequently than legislative party group meetings, but national
party executive committees generally have authority to establish the
party line. There is frequently some overlap between members of party
executive committees and legislators, particularly among legislative
group leaders. Many parties also retain disciplinary bodies, composed
of national party leaders, that are authorized to impose sanctions on
legislators who break discipline on votes where a party line has been
established.
Among the partisan groups from which I have interviewed deputies
(see the Appendix for a list), in Bolivia, Colombia, Costa Rica, Ecuador,
El Salvador, Mexico, Nicaragua, Peru, and Venezuela, there is remarkable
consistency on how decisions are made. Unless consultation is sought
1
Generally, the connection between electoral and legislative parties is straightforward, but it
may not be. Party switching in the legislature between elections is common in some coun-
tries, particularly where legislators are elected on the basis of personal votes, where volatility
in party support is high, or both (Thames 2007). Brazil is notorious on both counts (Despo-
sato 2006b; Samuels 2000). Rules of procedure in many legislatures also require some
minimum membership level for registration of a party group, so parties that have insuffi-
cient numbers may be forced either to coalesce in the legislature or to forgo whatever
resources are allocated to groups.
2
Venezuela’s 1999 constitutional provision (Art. 201) prohibiting such constraints is unusual
in this respect.
25
from, or imposed by, the national party organization, the norm is that
decisions are made in party group meetings by majority rule. This applies
to the question of whether to require discipline (or, alternatively, to leave
a matter open to conscience) and, if so, what the party line should be.
In a few cases, a provisional position for a party group can be set by the
group’s leaders themselves, although in cases where such an initiative
prompts dissent the fallback is to deliberate and vote within the party
group. According to Hugo Carvajal, Movimiento Izquierdista Revolucio-
nario deputy and former president of Bolivia’s Chamber of Deputies,
The bancada decides [its position] depending on the parliamentary rhythm; the
parliament has rhythms. The consultation sometimes gets only as far as a bancada
leader, and he defines a position and then transmits it to the group – we could say he
‘‘socializes’’ with the members – this decision that he’s already made and has adop-
ted in the name of the collegiate body. Sometimes this produces short circuits in the
members’ reaction.
In the case of such short circuits, the remedy is deliberation and a vote
within the bancada.
2.2.2. Discipline
Across the overwhelming majority of parties in the countries where I
conducted interviews, legislators reported that most votes are matters of
discipline. Without estimating precise rates of discipline, Salvadorans con-
curred that open votes are rare events (A. Alvarenga and Duch interviews).
Similarly, Carvajal estimated party-line voting 85 to 90 percent of the time
in Bolivia’s MIR, Guillermo Landazuri estimated the rate to be 90 percent
within Ecuador’s Izquierda Democratica, and Xavier Neira reckoned it
higher within that country’s Social Christian Party (Carvajal, Landazuri,
and Neira interviews). In Costa Rica, open votes appear to be slightly
more common, as do breaches of the party line on disciplined votes. The
chief of staff to the Costa Rican Assembly’s mesa directiva, Eladio Gonzalez
(interview), estimated that across all parties 80 to 85 percent of votes are
subject to discipline among all parties.
The most noteworthy exception to regular decision making at the level
of the party group was in Colombia, where legislators from both major
parties, the Liberals and Conservatives, in both chambers reported that
there are infrequent group meetings and that the majority of votes are left
open, without any established party decision or direction on how to vote
26
Legislative Parties and Discipline in Latin America
(Acosta, Devia, Garcı́a, Gómez Gallo, Navarro, and Holguı́n interviews).

As Hernán Andrade (interview) put it,
Because . . . practically all the members of our party, as in all of the parties, were
elected thanks to our own efforts, there’s no feeling of unity and there’s no mech-
anism in the bancadas that allows for prior, coordinated decision making, except in
exceptional cases. . . . Each of us is his own party. Each of us gets here due to our
own effort, with our own financing, with our own friends, without any clear ideol-
ogy – most of the time hiding the party we come from. There’s no channel that leads
to the bancada, there’s no partisan attachment.3
In explaining the sources of intrapartisan divisions, interview subjects

concurred that cohesiveness tends to be greater in smaller groups, where
there is more homogeneity of opinion, but that this is offset by economies
of scale that larger groups enjoy in providing benefits that induce loyalty
among legislators. Benefits range from physical resources, like offices and
staff, to committee assignments, to favorable treatment for private member
bills and budgetary funds for individual legislators’ chosen projects (de la
Cruz, Hernández, and Hurtado interviews).
2.2.3. How and When Sanctions Are Imposed

Legislators from all parties could cite cases of indiscipline, and they offered
various accounts about how, and how effectively, parties respond. Consis-
tent with academic accounts, pre–Hugo Chavez Venezuela appears to have
produced nearly airtight discipline across the party spectrum (Coppedge
1994). Combellas (interview) affirmed that breaches of party discipline in
legislative voting were rare in all parties, and that every instance – in state
assemblies as well as at the national level – triggered expulsion by the
national party organization. He noted, however, that the ‘‘conscience’’
provision in the 1999 Constitution (Art. 201) may provide judicial pro-
tection for undisciplined politicians.
Legislators in every country except Colombia acknowledged the exis-
tence of procedures to provide for expulsion on grounds of indiscipline, but
most emphasized that party leaders prefer to induce loyalty by other means,
if possible, particularly ones that are less public and dramatic. Vicente
Albornoz (interview), of Ecuador’s Democracia Popular, for example,
3
Colombia subsequently reformed the extremely individualistic electoral system to which
Andrade refers in this comment, replacing it with one that retains substantial individualism
but increases somewhat the extent to which copartisans’ electoral fates are intertwined.
27
noted that expulsion is used only if a breach of discipline is pivotal to the

outcome of a vote, citing a recent tax increase in which votes by renegade
deputies had turned the outcome. Nonpivotal defections, on the other
hand, tend to be tolerated without formal sanction.
In some countries, including Bolivia, El Salvador, and Nicaragua, elec-
toral party lists assign both a primary legislator (propietario) and a substitute
(suplente) to each legislative seat. When the propietario is unwilling to sup-
port the party line but willing to recuse himself from a vote, parties sum-
mon the corresponding suplente (A. Alvarenga, Samper, and Sanchez de
Lozada interviews). Only in Mexico, where I interviewed just the president
of the lower house and his staff, was there any reluctance to discuss mech-
anisms by which party discipline is enforced.
The most common theme running through accounts of party discipline,
by legislators across parties and systems, was control over career prospects.
Not surprisingly, all legislators are acutely aware of party control of this
critical resource. According to Alexis Sibaja (interview), Costa Rica’s mi-
nority party leader,
There is party discipline because political careers in Costa Rica are partisan. My
future is in Liberación Nacional (PLN), not outside it. I am disciplined every day
because I’m always interested in advancing within the PLN. . . . Desertions on
important matters are judged harshly by party militants and supporters. Those who
have deserted the party line in the past have effectively been retired from politics
because the party is very strict.
Academic accounts, as well as those of other interview subjects (Vargas

and Vargas Pagán interviews), suggest that Sibaja overstates the inviolabil-
ity of party discipline in Costa Rica, but he is unambiguous about the
source of what discipline exists. To the extent that legislators are ambitious
for political careers and parties can control access to these careers, parties
can induce legislative discipline. In August 2000 Mónica Baltodano (in-
terview) described hardball politics within Nicaragua’s Frente Sandinista
(FSLN) over her breach of discipline two months earlier on an electoral
reform bill on which the Frente had agreed to a compromise with the
governing Liberals:
Party rules say that on issues of national importance, the party organs decide and
the bancada is subordinate to these decisions. . . . The Sandinista national assembly
decided to go ahead with this reform, and they gave us the chance to express our
points of view. Afterward, we [four FSLN legislators] broke discipline. Then,
according to the statutes, we could have been sanctioned with expulsion or other
measures. This wasn’t convenient to them, politically. So they ruled that whoever
28
Partisan Representation Falling Short
did not accept party decisions could not aspire to electoral posts. Everyone knew I
wanted to run for mayor of Managua, and this way I couldn’t be nominated. It’s
almost certain that they won’t permit me to run for reelection as a deputy either.
And they took other measures. I was vice president of the Assembly’s executive
committee, and they took that away, and they won’t let me chair any committees.
Baltodano correctly anticipated continued conflicts with party leaders

over her aspirations. In December, facing public rebukes from party leaders
for failing to support their chosen nominee for the 2001 presidential elec-
tion, Baltodano noted that if she were in violation of party protocol, the
FSLN was bound by its own statutes to expel her, ‘‘But they have not done
that; therefore they cannot trample on any of my rights’’ (Latin America
Data Base 2001). Although the party’s executive committee stopped short
of expulsion, it issued a statement formally barring Baltodano from nom-
ination for reelection, citing as the reason her vote in the Assembly against
the electoral reform law.
The most consistent theme in the interview responses regarding sources
of party discipline is that parties have sanctioning mechanisms on the
books, but except for unusual circumstances, less formal measures serve
to induce discipline by appealing to legislators’ career ambitions.
2.3. Trouble in Paradise: Partisan Representation Falling Short
2.3.1. The Costs of Collective Reputations

Legislative accountability is complicated by a basic tension between party
discipline and individual responsiveness. The problem is that unified collec-
tive action by its legislators is necessary for a party to pursue its collective
goals, whether the goal is to implement policy or to capture state resources,
but the discipline required for effective collective action can undermine in-
dividual legislators’ responsiveness to their constituents. There is a trade-off
between demanding that legislators follow the party line and allowing them
flexibility to cultivate support by responding to diverse interests.
Parliamentary systems of government are widely held to resolve the
tension between collective and individual accountability in favor of the
former. So long as legislators value the stream of future benefits associated
with sustaining the government, the collective responsibility of governing
entities from cabinets to assemblies places a premium on party and co-
alition discipline over individualism (Cox 1987; Huber 1996; Diermeier
and Feddersen 1998).
29
Whether in parliamentary or presidential systems, coherent party labels

and the legislative efficiency achieved by discipline are both public goods
shared among supporters of the party that provides for them. Breaching
discipline may serve the particular interests of a given legislator’s support-
ers, but it degrades the public good shared by copartisans. To wit, the
fundamental dilemma of party discipline, and the root of the tension be-
tween discipline and individual responsiveness, is that ‘‘My own discipline
to the party is in my interest only when it is in my interest, but your
discipline is in my interest generally.’’
In Chapter 6, I present cross-national evidence confirming that govern-
ing parties in parliamentary systems are more unified in legislative voting
than those in presidential systems. For now, I simply note that even observ-
ers of European parliamentary systems document that breaches of disci-
pline occur and are products of demands for direct responsiveness by
representatives to their electoral constituencies when these run counter
to collective partisan objectives (Lanfranchi and Luchi 1999; Whitely
and Seyd 1999). Studies of European public opinion have noted a general
decline in trust in political parties over the past few decades (Pharr, Put-
nam, and Dalton 2000). Scarrow (2001a) attributes a rise in provisions for
direct election of executives and local officials, as well as in initiatives and
referenda, in Organisation for Economic Co-operation and Development
(OECD) countries to citizens’ decreasing trust in parties. Similar trends are
evident among the presidential democracies of the Americas. Although
public opinion survey data do not extend back so far as in Europe, the
Latinobarómetro (2003) annual survey picks up declining trust in political
parties regionwide from the mid-1990s to the early years of this decade.
Barczak (2001) documents an increase in provisions for, and the use of,
direct democracy throughout Latin America, which she attributes to wide-
spread popular dissatisfaction with political parties and the promise to in-
crease the responsiveness of political institutions to popular demands.
In short, there is a widely acknowledged tension between collective
and individualized accountability, and some scholars have noted signs
of dissatisfaction with party-dominated representation. For the most part,
however, studies of legislatures in presidential systems, and in Latin Amer-
ica in particular, have been critical of individualized representation, dem-
onstrating a normative bent toward strong parties capable of coordinating
legislative actions. This position reflects a preoccupation among Latin
Americanists about the marginalization of legislatures by powerful execu-
tives, but it is also rooted in a preference for collective partisan
30
accountability behind a broad program of government (Linz 1994; Valen-

zuela 1994). A central idea here, as in the literature on accountability in
parliamentary systems, is that strength in these aspects is necessary for
parties to be able to offer citizens coherent choices over policy and, in turn,
be judged by citizens in elections on the basis of past performance and the
credibility and appeal of their promises for the future (Luna and Zechmeis-
ter 2005; Rosas 2005).
2.3.2. Political Reform and Individual Accountability: Mixed-Member

Electoral Systems
Political reformers appear little concerned with currents in academic
opinion – in this case, with the normative emphasis on strong parties.
Throughout Latin America, a number of reform measures in recent years
have aimed to disconnect legislators from national party leadership when
this conflicts with responsiveness to local constituencies. A prime example
is the adoption in Bolivia, Guatemala, Panama, Venezuela, and Mexico
in the past two decades of mixed electoral systems, combining single-
member districts (SMDs) with proportional representation (PR) in over-
arching districts.4 The explicit goal of such reforms is most often to
tighten the local constituent-legislator bond, even at the expense of dis-
cipline among national parties.5 As part of an effort to resuscitate support
for a discredited party system in the early 1990s, for example, the Pres-
ident’s Commission on State Reform (COPRE) in Venezuela advocated
the shift from closed-list PR to SMD-PR elections on the grounds that
the previous system
strengthened the party line, which is defined by the top party leaders and the
tribunals of discipline responsible for its application. As a result, the legislators vote
as the party dictates without attending to the demands and interests of voters in
their regions . . . [whereas legislators elected under the proposed SMDs] ought to
act in the interests of their electors, ought to attend to their demands, ought to
respond to their mail, and will have to explain to their electors why they vote as they
do in the deliberative body. (Rachadell 1991:207–8)
4
Ecuador, meanwhile, combines personal voting in two-seat districts with closed-list pro-
portional representation in an upper tier.
5
Mexico used a straight SMD plurality until the 1970s, approaching the mixed system from
the opposite direction, adding PR seats gradually from the late 1970s to the early 1990s, to
allow for minority-party representation while maintaining the advantage that SMD plural-
ity tends to provide for the largest party (Molinar Horcacitas and Weldon 2001).
31
The same motivation spurred the shift from pure closed-list PR election
to SMD-PR in Bolivia in 1994 where the plummeting stature of political
parties, evident in street protests as well as opinion polls, was understood as
a demand from voters ‘‘that deputies should be known and acknowledged
representatives of their constituencies and not anonymous representatives
of party leaders. Direct connections between deputies and voters would
therefore enhance the legitimacy and representativeness of the parliament,
making possible the responsiveness and accountability of deputies to their
constituencies’’ (Mayorga 2001). Precisely the same arguments were
made by advocates of a proposal for SMD-PR in Costa Rica in 2000
(Sibaja interview).
Whether these reforms deliver enhanced individualistic representation
and accountability remains an open question. In Bolivia, an early review of
the local responsiveness of SMD deputies (called uninominales) found only
modest improvements (Culver and Ferrufino 2000). In Venezuela, elec-
toral reform in the late 1980s proved insufficient to salvage public support
for the traditional parties and avert their complete collapse in the 1990s.
Yet the COPRE’s recommended SMD-PR format survived even President
Chavez’s constitutional overhaul of 1999; indeed, the champions of Cha-
vez’s reforms echoed the COPRE’s calls for strengthened electoral ties to
local constituencies (Combellas, Tarek Saab, and Fernández interviews).
Finally, although the empirical focus of this book is primarily Latin
America, it is worth noting that the arguments made in that region resonate
as well among SMD-PR advocates in Europe and elsewhere. In the past
decade and a half, Italy, Japan, New Zealand, the Philippines, Russia, and
Ukraine have adopted mixed SMD-PR electoral systems.6 Richard Katz
describes the Italian electoral reform of 1994 as motivated by popular
demands for alternation in government and for ‘‘direct accountability of
individual members of parliament to their electors. There was a desire to
free the electorate from the confines of party labels and ideologies, and to
allow electors to take into account the character, qualifications, and per-
formance in office of individual candidates when casting their votes’’ (Katz
1994:103). Historically, the ‘‘mother of mixed systems,’’ in Germany, was
distinguished by its advocates from the prewar Weimar system of closed-
list PR by its virtue of strengthening the connection between voters and
individual representatives. As noted by Susan Scarrow (2001b:63),
6
Germany has used such a system since 1949.
32
German advocates of mixed-member rules argued that such rules would ‘‘person-
alize’’ voters’ choices by letting them choose individual representatives from small
districts – indeed, Germans still refer to their system as being an example of ‘‘per-
sonalized PR,’’ a label that is meant to distinguish it from proportional systems that
lack a nominal tier.
2.3.3. Venezuela’s Constitution of 1999

Perhaps the most controversial institutional reform in recent decades in
Latin America was the adoption, at the behest of President Hugo Chavez,
of a new constitution for Venezuela in 1999. Chavez’s critics decry him as
an autocrat who has systematically removed or enfeebled any meaningful
checks on his own power (McCoy 1999; Gunson 2006). His admirers tout
his reforms as improving the quality of democracy in a system long dom-
inated by an entrenched and unaccountable oligarchy of parties – a ‘‘party-
archy.’’ Without attempting to arbitrate this debate, which promises to
endure well beyond Chavez’s presidency, I note here only that the
1999 constitution includes a number of new measures aimed specifically
to foster personal responsibility by legislators to their district
constituencies.
There is a four-year residency requirement for eligibility to run for the
legislature from any given district or state, designed to ensure that repre-
sentatives know firsthand the needs and preferences of district voters (Art.
188). Legislators are obliged to ‘‘render accounts’’ of their activities each
year in public forums (rendiciones de cuentas) in their districts and to explain
and defend their behavior and their votes (Arts. 197, 199). All legislative
votes are explicitly deemed matters of individual conscience for represen-
tatives, rather than matters of partisan obligation (Art. 201). Finally, all
elected officials are subject to recall elections, which can be initiated by
a petition of 10 percent of the voters in their districts (Art. 72, 197).
To the extent that forcing legislators to render accounts to their
districts produces additional information for constituents about legisla-
tors’ actions, its connection to the idea of accountability established in
Chapter 1 is straightforward. More broadly, game-theoretical analysis
suggests that requiring individual representatives to explain votes
increases the efficiency of electoral punishment for legislators otherwise
inclined to ignore constituents’ wishes and, in doing so, enhances respon-
siveness at the individual level (Austen-Smith 1994). All these anticipated
effects were articulated – albeit, without the game theory – by Venezuelan
legislators in interviews. Ricardo Combellas, a constituent assembly
33
delegate and opponent of President Chavez, describes the motivation

behind the reforms:
We wanted to eliminate partyarchy – not to eliminate it constitutionally, but in

terms of norms, for the representative to respond more directly to the wants and
needs of his constituents. His responsibility in parliament is personal – the Con-
stitution says so – not to respond to a party but to his constituents. We estab-
lished a rendering of accounts that didn’t exist before . . . [and] a vote of
conscience that wasn’t there either. . . . [In the past], the parties overwhelmed
their representatives. They imposed the line, imposed the vote, imposed atti-
tudes. We have tried to relax this and create a more fluid relationship between
legislators and their constituents. Besides, a legislator now has to have lived at
least the past four years in the region where he is elected. And we have recall
elections. All this is to say that there are innovative constitutional reforms, very
different from what we had before, but that we don’t know how they’re going to
work. That much will require a cultural change, but what we did with the Con-
stitution was important.
Referring to the same set of provisions, chavista constitutional delegate

and later deputy William Tarek Saab is even more unrestrained in his in-
terview: ‘‘A big space is opened where the parties used to have complete
control, and power is completely realigned. I think that we have put orga-
nized society above the parties – that the organized people, the organized
popular movement will have a chance now because these constitutional
measures give them a chance.’’ Whatever effect, if any, these provisions
have on legislative representation in Venezuela in the long run, Chavez
supporters, and at least some skeptics, justify their intent as increasing the
personal accountability of politicians, even if this loosens the bonds of party.
2.3.4. Citizen Demand for Legislative Individualism

Measuring overall citizen opinion about the proper balance between party
control and individual legislative responsiveness is difficult. Nevertheless,
some surveys tap into the issue, as do data from elections in countries where
ballots offer voters the choice of casting a party or an individual preference
vote. Again, these eclectic sources support the proposition that there is
a deeper reservoir of public support for individual-level accountability than
is reflected in most of the academic literature.
In a rare public opinion poll addressing the matter of legislative
individualism directly, 1,505 Chileans were asked in 2007 whether, in gen-
eral, deputies and senators in Congress ought to vote according to their
34
own preferences or the preferences of their parties. Nearly twice as

many respondents wanted legislators to vote their own preferences
rather than with their parties – 50 to 28 percent (Centro de Estudios
Públicos 2007).7
A 2006 survey in Bolivia tapped into the same sentiment, albeit indi-
rectly, during the lead-up to elections for a constituent assembly, which was
widely expected to declare itself sovereign and appropriate the powers of
the existing legislature. In this context, 3,013 Bolivians were asked how
members of such an assembly should be elected: by political parties, by
citizen groups, by indigenous groups, by labor unions, by municipal com-
mittees, in single-member districts, or none of the above.8 The two alter-
natives with which survey respondents were most familiar were by parties
and in SMDs, because Bolivian democracy used closed-party-list PR elec-
tions up through the mid-1990s, and a mixed-member system combining
SMDs with closed-list PR thereafter. Of these two options, SMDs are
widely associated with individual-level accountability and election by par-
ties with collective accountability. More than twice as many survey
respondents preferred election in SMDs to election by political parties –
19 to 7 percent (Seligson et al. 2007:106–7).9
Finally, consider that, in the current decade, both the Dominican Re-
public and Colombia adopted ballots that allow voters the option of casting
a preference vote for an individual candidate or endorsing a single slate
presented by their party.10 Given the choice, voters overwhelmingly used
the individual-candidate preference vote. In the Dominican Republic’s
2006 election, its first with the preference vote, 80 percent of voters exer-
cised that option (Morgan, Espinal, and Seligson 2006:135). Colombian
ballots first offered the preference vote in 2005, and 80 percent of voters
used it there as well (Shugart, Moreno, and Fajardo 2006, table 7.9). In
Brazil, which has offered the list-versus-preference option longer, between
82 and 92 percent of voters exercised the preference vote option in elections
from 1990 to 2002 (Nicolau 2007:108).
7
Another 11 percent each said it depended on the circumstances, or offered no opinion.
8
The menu is, admittedly, a bit ambiguous, in that the details of selection within these
groups were not spelled out.
9
The plurality choice was by citizen groups (34%), with other responses including munic-
ipal committees (11%), indigenous groups (6%), unions (5%), and no opinion (15%).
10
The Dominican Republic previously used closed-list PR. Colombian lists were also closed,
although each party was allowed to present multiple lists, injecting substantial individual-
ism into Colombian elections.
35
2.3.5. The Move toward Individualism

The reforms discussed here were developed independently, and the opin-
ions and behaviors cited are drawn from various legislative environments,
but a common thread running through them is the stated intention to
strengthen the accountability of individual legislators to voters. At least
in their rhetoric, contemporary Latin American political reformers are
critical of legislative party discipline because it conflicts with the individual
accountability that they endorse and that citizens apparently desire. In the
next chapter, I discuss a reform I regard as integral to any shift toward
individual legislative accountability – recorded voting. For now, I turn to
some survey evidence that the expressed preference for greater legislative
individualism is widespread among legislators themselves.
2.4. The View from the Chamber

During the late 1990s, and again during the first half of this decade, the
Proyecto de Elites Latinoamericanas project conducted surveys of legisla-
tors throughout Latin America on an array of issues, including the princi-
pals to whom they are responsive and their disposition toward party
leaders. Some of the questions were repeated across the two rounds of
surveys, and attitudes toward party leadership and discipline in these cases
show remarkable stability over time (Alcántara 1994–2000; Proyecto Élites
Parlamentarias Latinoamericanas 2006). In most cases, the team was able
to get responses from well over half the membership of the lower chamber
(or only chamber, in unicameral systems) and across the full range of par-
ties. Of course, one must take into consideration that the survey respond-
ents, the legislators themselves, might answer according to what they
regard to be norms of acceptable behavior. To put it less delicately, survey
responses could be self-serving. Nevertheless, there are good reasons to
expect that the surveys contain useful information, especially for the pur-
poses of comparison across countries.
Most important, the surveys are anonymous, so from a legislator’s per-
spective there is nothing to be gained from self-promotion. Contrast this
with the legislator interviews on which I draw throughout this book. I in-
troduced the interviews purely for the purposes of academic research, but I
did not present anonymity as the default condition, and almost no interview
subject requested it. Subjects frequently described their own actions as based
on personal ambition, political deal making, and compromise of principles
36
The View from the Chamber
(not to mention how they depict the machinations of their colleagues). If

interviewees are willing to portray their behavior in a manner that might be
regarded as unseemly despite the lack of anonymity, then it is reasonable to
expect a greater level of candor in the surveys.
The next three figures present the relative influence of three potentially
important principals – national party leaders, voters in their district, and
the government – on the decisions legislators make. Respondents were
asked whether they take the opinion of each principal into consideration
‘‘a lot,’’ ‘‘somewhat,’’ ‘‘a little,’’ or ‘‘not at all’’ when making political
decisions. In the interest of simplicity, Figure 2.1 presents the results with
respect to our universal principals, party leaders, with the combined
percentage of ‘‘a little’’ and ‘‘not at all’’ responses subtracted from the
percentage of ‘‘a lot’’ and ‘‘somewhat’’ responses.
The first thing to note is that a plurality of legislators in every country,
and big majorities in most, acknowledge paying substantial attention
to party leaders. Note that if 70 percent say ‘‘a lot’’ and ‘‘somewhat’’
while 30 percent say ‘‘a little’’ and ‘‘not at all,’’ the net score will still be
40 percent, as in Chile, for example. Venezuelan and Peruvian respond-
ents profess the most independence from party leaders, although this
could be at least in part a product of the specific cohorts of legislators
Figure 2.1 How much do you take the opinion of national party leaders into
consideration when making political decisions?
37
to which the surveys in these countries were administered. The Vene-

zuelan survey was administered in 2000, following directly on the ratifi-
cation of the new constitution, which repeatedly professes its
commitment to legislative independence from partisan control. The
Peruvian survey was administered in 2001, following the downfall of
President Alberto Fujimori’s government on corruption charges, in
a context where the principle of party discipline was tarred with the
brush of corruption (Carey 2003). On the whole, however, legislators
acknowledge substantial deference to party leaders. Colombian legislators
are generally regarded as dismissive of party leaders, as the interview
responses suggested, but even here only a little more than 30 percent say
they pay little or no attention to this principal, whereas twice as many pro-
fessed some allegiance.
On the other hand, in twelve of the thirteen countries for which I have
survey data, more legislators say they pay ‘‘some’’ rather than ‘‘a lot’’ of
attention to party leaders (not visible in Figure 2.1). Although legislators
take party leaders into consideration, they reserve a higher level of deference
for another principal, voters in their district, as indicated by Figure 2.2.
In every country surveyed, more legislators claimed to pay greater at-
tention to voters in their district than to any other factor when making
Figure 2.2 How much do you take the opinion of voters in your district into
38
The View from the Chamber
political decisions. Other potential sources of influence in the survey, but

not discussed in detail here, included national public opinion, voters from
within the legislator’s party, the media, and interest groups. According to
the surveys, none warranted such deference as district voters. This is note-
worthy, particularly because most of the legislators surveyed were elected
from closed party lists in which the direct link between district voters and
their representatives is tenuous at best.
Figure 2.3 shows the amount of attention legislators claim to pay to the
government in forming their decisions. Within each country, responses
separate according to whether the legislator’s party is allied with the
president or not, with the former indicating greater levels of deference to
this principal and the latter, lower levels. Even among presidents’ allies,
however, the levels of stated deference do not approach those legislators’
claim to voters in their districts.
As a final indication of this tendency, consider the survey question, ‘‘Do
you think the national party leadership should have more power over legis-
lators, less power, or maintain the same?’’ Figure 2.4 shows the percentage
responding ‘‘more’’ minus that responding ‘‘less,’’ such that negative
Figure 2.3 How much do you take the opinion of the government into
39
Figure 2.4 Do you think the national party leadership should have more power
over legislators or less?
numbers favor decreasing party leader control overall. In every country but
Colombia, more respondents said ‘‘less’’ than ‘‘more,’’ generally many times
more. In ten of fifteen countries, an outright majority of respondents in-
dicated a preference for less central party control. Across countries, the mean
level of support for increased party control is 13 percent, whereas the mean
support for decreased control is 56 percent.
According to the surveys, legislators claim they prefer more of their own
discretion, and less control from their parties, toward the expressed priority
of representing the interests of voters from their districts. All this may be
posturing, of course, if legislators for some reason felt obliged to dissemble
on the surveys. Even if the relative commitment professed for district voters
versus party leaders or presidents is not sincere, it indicates a professed
commitment to the sort of representation generally associated with legis-
lative individualism rather than collectivism.
2.5. The Shift toward Individual Accountability

In this chapter, I suggest that over the past decade various factors have
increased the sensitivity among legislators in Latin America to pressures
40
The Shift toward Individual Accountability
other than the demands of national party leaders. It is important to ac-

knowledge that even party leaders should not necessarily demand blind
responsiveness to the national command on the part of their troops. Total
failure by legislators to attend to local, sectoral, and even individual con-
stituent demands can leave national leaders sitting atop organizations with
no electoral support. This calculus by national leaders was responsible for
the adoption of mixed-member electoral systems in Venezuela and Bolivia
(Crisp and Rey 2001). National party leaders pursuing such a strategy may
parcel out reforms providing a modicum of individual flexibility while
retaining other powers and resources that ensure discipline. Thus, for
example, leaders of most Bolivian, Mexican, and Venezuelan parties have
maintained centralized control over candidate nominations, seriously lim-
iting the extent to which district pressures induce even SMD legislators to
buck party discipline (Mayorga 2001).
Despite these constraints, however, the overall trend is toward the ex-
posure of legislators to increasing pressures from sources in addition to
their parties. In the mixed-member systems, SMDs induce individual leg-
islative entrepreneurship and constituency service. Moreover, other elec-
toral reforms aimed at increasing voter discretion among candidates within
parties, such as preference voting within lists and primary elections for
candidate nominations, have been advanced in Costa Rica, Mexico, and
Venezuela. These reforms are expected to increase the willingness of legis-
lators to break party discipline in legislative voting (Kulischeck and Crisp
2001; Mayorga 2001; Weldon 2001; E. Vargas interview).
Even the Colombian electoral reform of 2003, which aimed to enhance
collective accountability by requiring that each party present a unified list of
candidates in each district rather than multiple lists, affords parties the op-
tion of presenting open or closed lists to voters. It is noteworthy that, in the
2006 congressional elections, the first to employ this new system, more than
80 percent of parties competing opted to present open lists and that more
than 80 percent of voters seized the opportunity to cast a preference vote
for an individual candidate rather than simply to indicate a choice for a par-
ty’s list as a whole (Shugart, Moreno, and Fajardo 2006). In short, among
reformers of late – and among voters – there is a strong attraction toward
individual accountability as a normative priority in legislative elections.
Have institutional reforms enhanced legislative individualism in Latin
America? This is a hard question to answer. The Proyecto de Elites Latino-
americanas data provide snapshots of legislative attitudes, but we lack
survey data over long time periods. In taking stock of specific reforms,
41
one confronts similar challenges. The 1999 Venezuelan Constitution is

still relatively young, and Venezuelan politics during its first few years
has been punctuated with government crises (Gunson 2006; Pérez-Liñán
2007). With respect to Latin America’s mixed electoral systems, early case
studies suggest some impact. Even before Chavez’s overhaul of the Vene-
zuelan Constitution, SMD-PR elections generated legislators with distinc-
tive perceptions of accountability. A majority of SMD deputies surveyed in
1997 claimed that citizens’ votes are based on the personal characteristics of
candidates, whereas more than 90 percent of PR list deputies contended
that voters ‘‘think of politics in terms of parties’’ (Kulischek and Crisp
2001). In Bolivia, similarly, one observer judged the mixed system to have
‘‘produced two classes of deputies and two different parliamentary roles.
Overall, a trend toward locality-centered politics, constituency-serving ‘re-
tail politics’ (and perhaps also corporative politics) has been strengthened at
the expense of national politics’’ (Mayorga 2001).
Systematic cross-national comparisons are more difficult, and I do not
offer a definitive judgment on the reforms discussed in this chapter and
their effect on individualism. In Chapters 5 and 6, I propose various in-
dices of party unity that measure the willingness of individual legislators
to buck party unity and evaluate the factors that generate high levels of
unity or, conversely, more legislative individualism. The indices are based
on legislative votes recorded at the level of the individual legislators.
But the entire project of studying individual voting behavior confronts
a fundamental and serious problem – namely, that in many legislatures
recorded votes are scarce or unavailable altogether. In the next chapters,
therefore, I explore the prior phenomenon of recorded legislative votes
themselves, describing where they exist, where they do not, why, and what
the practice of recording legislative votes implies for legislative account-
ability at the individual level.
42
3
The Supply of Visible Votes
Let us make public the names of those who voted in favor, so our children will know
whom they should curse.
Russian legislator Yuli Rybakov, on a proposal to accept nuclear waste from other countries in
exchange for cash (National Public Radio 2001)
3.1. Visible Votes and Accountability
3.1.1. Questions about Visible Votes

Whatever else voters in the United States know or do not know, they can
count on being alerted as to whom they should curse for any decision
Congress makes. Interest groups publish widely cited ‘‘report cards’’
based on legislative voting records, challengers comb through their in-
cumbent opponents’ records, and incumbents whose voting records are
out of sync with their districts’ interests pay an electoral price (Canes-
Wrone, Brady, and Cogan 2002). It is sometimes held that elected rep-
resentatives generally operate according to a calculus familiar to U.S.
legislators. In her cross-national study of corruption, for example, Susan
Rose-Ackerman (1999:127) offers as axiomatic that, ‘‘If politicians vote
against the interests of their constituents, they can expect to suffer at the
polls.’’ But is this true? In many legislatures, who voted for and against
a given proposal is almost never revealed, and proposals to record votes at
all are contentious.
The conditions that foster, or undermine, political accountability are in-
creasingly central to students of comparative democracies (Adserá, Boix, and
Payne 2003; Cleary and Stokes 2006; Johnson and Crisp 2003; Stokes 2001).
43
The broad question motivating this chapter, and the next, is whether the
information necessary to make individual accountability possible is available
from legislatures. Why tie the transparency of information about legislative
votes so tightly to individual but not collective accountability? This chapter
argues that the responsiveness of individual legislators depends heavily on
which of their potential principals can easily monitor their actions. Party
leaders can monitor voting in the legislature relatively easily, regardless of
whether the votes are published and transparent. A lack of transparency,
therefore, fosters asymmetry between party leaders and constituents. Voting
transparency is less relevant to citizen monitoring of parties as wholes. Even
when individual voting records are unavailable, those outside the legislature
can get access to information about the policy positions and actions of major
collective actors – parties, unions, business associations – at relatively low cost
through newspapers and broadcast media. This chapter emphasizes the con-
nection between transparency and individual accountability, contending that,
when votes are not readily visible, the costs of information about individual
legislative behavior are prohibitive for legislative outsiders.
To make the case, I address a number of specific questions: What in-
formation about votes is available? What factors generate change in the
revealed information about votes? What conditions are necessary for vot-
ing records to be politically salient? What effect does public voting have on
the relationship between individual representatives and their parties? As
the chapter discusses, the amount of information revealed about votes
varies tremendously across legislatures. Vote records are potentially salient
in all legislatures, although more in single-member-district than multi-
member-district electoral systems. Lawmakers disagree on whether public
voting is desirable, with those who control the agenda generally opposed.
Pressures from opposition and dissident lawmakers and from pro-
transparency actors outside legislatures, as well as technological advances,
all encourage publication of legislative voting records. Finally, there is
a tension between public voting by individual legislators and disciplined
voting among political parties.
3.1.2. Looking for Visible Votes in the Americas

I bring to bear on these issues evidence from Latin American legislatures, as
well as observations about the genesis of public voting in the United States.
The country cases included in this examination of legislative transparency
are described in Table 3.1.
44
Visible Votes and Accountability
Table 3.1. Data Included in Analysis of Legislative Transparency
Country Data on the Number Archival and Interview Archival Data Based
of Visible Votes Data Based on on Research
Field Work Assistance and
Correspondence
Argentina X X
Bolivia X X X
Brazil X X
Chile X X
Colombia X X X
Costa Rica X X
Ecuador X X
El Salvador X X
Guatemala X X
Mexico X X X
Nicaragua X X
Panama X X
Peru X X
United States X X
Uruguay X
Venezuela X X X
What are the implications of examining this particular set of cases? First,
all these cases are presidential, and together they constitute most of the
pure presidential democracies in the world. In these regimes, the separation
of powers allows legislative performance to be evaluated independently
from executive performance (Cox 1987; Diermeier and Feddersen 1998;
Samuels 2004). Because legislative votes on individual policy proposals do
not directly affect the survival of the executive in office, as they may under
parliamentary and hybrid regimes, accountability for individual voting deci-
sions ought to be more pronounced under presidentialism. If public voting is
politically salient anywhere, then it should be so in the Americas. In this
sense, the countries examined in Chapters 3 and 4 may be biased toward high
demand for legislative transparency. Amplified demand could facilitate the
job of detecting conditions that encourage transparency – for example, by
heightening sensitivity to whether particular electoral rules or procedural
rules favoring the opposition favor transparency of individual legislators’
behavior. If this is the case, it may be that in parliamentary and hybrid
regimes the effect of similar conditions on transparency should be muted.
45
The exclusive concentration on presidential regimes in the next two

chapters is largely a product of resource constraints.1 As an initial exami-
nation of legislative voting transparency, it focuses on the ‘‘most likely
suspects.’’ The U.S. case is a benchmark in assessing the importance of
public voting because that is where the practice is most firmly established
and most widely recognized to be politically consequential. In Latin Amer-
ica, demands for individual-level legislative accountability have increased
in recent years, and the availability of reliable electronic voting equipment
has dramatically reduced the logistical barriers to public voting (Barczack
2001; Mayorga 2001; Rachadell 1991). The interview and archival evidence
presented in these chapters suggests that voting transparency can be central
to legislative accountability – at least under presidentialism, an institutional
format expected to favor scrutiny of individual legislators – and the evi-
dence sheds light on the specific conditions that foster such transparency.
3.1.3. Why Focus on Legislative Votes?

This book rests on the premise that voting records provide important
information about the actions of parties and of individual legislators
and that it is worthwhile to compare this information cross-nationally.
Is this a sensible way to proceed? It is important to acknowledge that
voting on the floor is far from the only consequential action that takes
place in most legislatures. Much of the policymaking and bargaining
action in any legislature takes place before proposals reach the voting
stage – in public pronouncements and debate, in legislative committees
and party caucuses, or during negotiations between executive and legis-
lative actors or between party leaders and rank-and-file legislators. Tran-
scripts of legislators’ speeches and floor debates, for example, which
generally fill volumes of published records (variously called Boletines,
Gacetas, Hansards, Records, Proceedings, etc.) from assemblies around the
world, are a staple of legislative behavior. In a remarkable recent innova-
tion, Quinn et al. (2006) have developed software to scan text records of
legislative debate for rhetorical content that can then be used to locate
speakers’ ideological positions in a multidimensional ideological space in
1
The discussion in Section 5.4 of data included in the analysis of recorded voting in Chapters
5 and 6, which extends beyond the Americas and beyond pure presidential regimes, elab-
orates further on how research resources affect access to data on legislative voting, and with
what implications.
46
Visible Votes and Accountability
a manner analogous to methods based on recorded vote data that are

common currency in scholarship on the U.S. Congress (Poole and
Rosenthal 1985, 1997). An advantage of the Quinn et al. method is that
it does not require access to voting records, so it can be applied anywhere
politicians go on record doing what they do best (or most, at any rate),
which is to speak. Once applied comparatively, this method will provide
a fine-grained ideological map of legislatures. Yet it will not indicate
directly whether legislative copartisans act in such a way as to pass the
policies they profess to support. This is not just because legislators might
say one thing but do another when it comes time to vote; it is also because
rhetorical ideological proximity does not identify the dividing lines be-
tween support and opposition for a given proposal, much less whether
breaches in voting unity cause parties or coalitions to lose votes they
might otherwise have won.2
It is also important to acknowledge that there are good reasons for
caution in interpreting analyses of recorded votes as complete portraits of
legislators’ support for the proposals with which they are confronted, and
for the authors of those initiatives. Consider, for example, the prospect
of drawing inferences about the legislative effectiveness of executives
based on the success rate of executive-sponsored bills in legislative votes
(Figueiredo and Limongi 2000; Siavelis 2000; Cheibub, Przeworski, and
Saiegh 2004; Calvo 2007). Ames (2002) sounds a cautionary note by
documenting the incidence of executive policy initiatives in Brazil that
are delayed, are modified, or die outright before ever reaching the point of
a recorded vote. Such action is clearly critical to the legislative process but
is effectively invisible to analyses that are limited to recorded votes taken
on the legislative floor.
The general point that much important legislative action never shows up
in floor votes is indisputable, but it does not imply that the information
contained in floor votes is unimportant. Indeed, many crucial policy
choices in most democracies – like those on annual budgets, the appointment
of government officials, or international agreements – are constitutionally
required to be voted on the floor of the legislature.
A related rationale for skepticism about the relevance of floor votes is
that, if party leaders have good information about legislators’ preferences,
2
This is not meant as a criticism of rhetorical ideal point estimation, which is a ground-
breaking innovation, but merely to note that it is a distinct, and complementary, endeavor
from measuring voting unity.
47
voting outcomes themselves may be foregone conclusions. In the extreme

case, the information available in floor votes might not be inconsequential
but rather unrepresentative of what goes on within legislatures. In a major-
ity party or coalition, for example, the leadership may dictate the legislative
agenda and have perfect information about how all legislators will vote on
any proposal. If those leaders hate to lose, they may allow no votes on any
proposals where their side will lose. The votes we observe, then, will tell us
something about where legislators stand, and probably about what they
fight for and against in other aspects of their legislative duties, but the
picture will be incomplete and biased.
It is undeniable that parties and governing coalitions – and particularly
their leadership – work constantly to turn the legislative agenda to their
advantage. When the task at hand is to infer from floor votes the prefer-
ences of legislators or the unity of parties and coalitions, the implications
of agenda control should always be kept in mind. Yet the potential dis-
tortions in the information contained in floor votes implied by agenda
control should be examined rather than assumed. Leadership information
about legislators’ preferences is never perfect. Legislative rules sometimes
allow qualified minorities to demand votes, and so challenge majority
agenda control. Some votes, as noted earlier, are mandatory. Control over
the agenda is generally less than absolute, and how it is distributed
depends on institutional and political factors in ways that can often be
theorized, measured, and tested. In subsequent chapters, I show that floor
voting is generally not just a string of faits accompli. Records almost always
include votes that are divisive within the assembly, and often within legisla-
tive parties and coalitions themselves, and vote outcomes turn on these
divisions.
For now, it is sufficient to note that floor voting is a critical procedural
element of all democratic legislatures. There are theoretical and empirical
reasons to expect that floor voting patterns can provide relevant informa-
tion about what it is that legislators value, and about how effectively they,
and the groups into which they organize themselves, pursue those values.
This presents us with a puzzle, however; in many legislatures, most of the
information contained in voting records is invisible to all but those present
for the votes themselves.
The remainder of this chapter examines how much information about
legislative votes is visible to those outside the legislature. First, I present
a typology of legislative voting methods according to whether they can be
monitored by observers inside the legislature, outside, both, or neither, and
48
Who Can Monitor Votes?
develop propositions regarding the politics of public voting based on these

conditions. Next, I review the historical trajectory of public legislative
voting, and its political salience, in the United States, and then turn atten-
tion to Latin America, surveying the extent of public voting, and the
process by which it was adopted in some recent cases. The empirical basis
for much of this chapter, and the next, is a series of interviews conducted by
the author with fifty-six legislators, party leaders, and legislative staff,
during 2000–1, in Bolivia, Colombia, Costa Rica, Ecuador, El Salvador,
Nicaragua, Peru, and Venezuela.
3.2. Who Can Monitor Votes?

Table 3.2 distinguishes voting procedures by the relative ability of actors
inside and outside the chamber to monitor individual legislators’ votes
Table 3.2 Monitoring Legislative Votes
Internal (e.g., party leaders)

No Yes
External (e.g., citizens)
No Secret voting Signal voting

U.S. House officer Latin America standard
elections, pre-1839 operating procedure
in lieu of electronic voting
Peru, at request of U.S. House voice votes

one-third of legislators,
or Panama at request
of one-half
Italian final passage
votes, pre-1988
Yes NA Public voting

U.S. Congress roll call and
teller votes
Latin America nominales
Latin America with

electronic voting when
records are made available
49
and illustrates the three relevant configurations, along with examples of

each type of procedure that is discussed in the text. Under secret voting,
legislators cast anonymous ballots such that the position of each is un-
known by any monitor. Under public voting, the position of each legisla-
tor (most commonly, ‘‘aye,’’ ‘‘nay,’’ ‘‘abstain,’’ or ‘‘not voting’’) is
generally published in some official journal of the legislature and often
available on a legislative Web site. Under both secret voting and public
voting, the ability of actors internal and external to the legislature to
monitor individual votes is symmetrical. Signal voting refers to proce-
dures by which individual legislators’ positions are visible to those phys-
ically present in the chamber, but no individual-level record is available
to outside actors, introducing asymmetry in the capacity of internal and
external actors to monitor individual votes. Most votes in most Latin
American legislatures – and almost all votes in many – are signal votes.
The mechanics generally involve hand raising (‘‘All in favor . . .’’) or
standing up to be counted.
An example from Argentina illustrates the difficulties of monitoring
legislator behavior when signal voting is used. On April 26, 2000, the
Argentine Senate approved sweeping reforms to the country’s labor code
in a series of signal votes. Subsequently, allegations raised in the press, then
by the administrative secretary of the Senate, held that the government of
new president Fernando de la Rua had bribed some senators for their votes,
triggering a scandal that prompted the resignation of Argentina’s vice
president and gravely wounded de la Rua’s presidency. Yet determining
which votes were allegedly bought proved impossible. Of fifty-nine sena-
tors present for the vote on final passage of the Labor Code, eleven spoke in
favor of it on the floor, four spoke against, and the rest left no trace
(Gonzalez Bertomeu 2004:39–40).
Senator José Luis Gioja was singled out as having allegedly betrayed
his Peronist copartisans in opposition to de la Rua and sold his affirmative
vote to the government, but there was no way to determine whether
Gioja’s vote was registered in support of the administration’s reform in
the first place. The video record of that session showed Gioja, during
what became known as the ‘‘grooming vote,’’ run his hand through his
hair, touch his face, adjust his glasses, then turn to talk to his colleague
(Clarıń 2004a, 2004b). In short, in a case where an explosive charge of vote
selling was at stake, signal voting made it impossible even to evaluate the
premise of the charge. Most signal votes are, of course, less quirky in their
execution and less consequential in their impact on policy and politics,
50
The U.S. Experience
but they produce official records equally inscrutable in terms of individual

accountability.
The rules of procedure in all Latin American legislatures include
provisions for public voting. These votes are usually called nominales, or
named votes, in which the roll is called and each legislator’s position is
recorded. Requirements for demanding a nominal range from requests by
a handful of legislators to petition by a majority of those present. Apart
from these thresholds, traditional nominal procedures are inevitably
time-consuming and procedurally costly. Electronic voting machines,
in contrast, generate individual-level voting records automatically, so
that, when they are used, the procedural costs of public voting plummets,
regardless of whether electronic votes are formally deemed nominales
(Congreso del Perú 1998).
3.3. The U.S. Experience
3.3.1. Voting Records as the Currency of Individual Accountability

I rely on the U.S. Congress as a point of reference for my examination of
recorded voting, not because it necessarily represents a normative ideal for
legislative organization, much less because it is empirically representative
of legislatures more generally, but rather because recorded voting is more
ubiquitous, and has been for much longer, than anywhere else. The sus-
tained centrality of voting records to U.S. politics allows for scholarly
examination of how visible votes have been perceived by various political
actors and how they are connected to legislative accountability. My brief
discussion of the United States highlights four points. First, public voting
encourages legislators to be responsive to constituent interests. Second,
this has been the case since the early days of the republic. Third, public
voting imposes a strain on party discipline by exposing representatives to
pressures from outside the legislature. Fourth, technological advances that
reduce the procedural costs of recording votes increase their salience as
tools for dissident and opposition legislators. These themes are echoed in
subsequent discussion of public voting in Latin America, although some-
what more faintly than in the United States, because in Latin America the
conditions for voting records to serve as tools of accountability are, and
have long been, less propitious.
The first point is uncontroversial. The centrality of voting records
to campaign strategies is apparent to any observer of U.S. legislative
51
politics. Incumbents try to avoid casting votes that potential challengers

could trot out as evidence that constituent interests have been betrayed.
The connection is also established in academic research, both qualitative
and quantitative (Erickson 1971; Fenno 1978). In Mayhew’s seminal
account of the electoral connection, roll call voting records are an essen-
tial component of legislators’ position-taking strategies (Mayhew
1974:69–73). In a recent analysis of recorded votes across four decades,
from 1956 to 1996, Canes-Wrone, Brady, and Cogan (2002) demonstrate
that U.S. House members whose floor votes prioritize the demands of
their parties over their constituents win lower shares of the popular vote
and face higher probability of defeat than do members whose votes are
more in line with the estimated preferences of voters in their districts.
The results suggest that constituents are aware of their representatives’
voting behavior and that electoral ambition induces responsiveness to
constituent preferences.
3.3.2. Punishment at the Polls

Modern campaigns and communications media facilitate dissemination
of information on voting but also raise the question whether legislative
politics in the United States has always been characterized by account-
ability of this sort. The historical record suggests that it has. Consider, for
example, the controversy surrounding the Compensation Act of 1816, in
which a Republican-controlled Congress voted to switch from per diem
compensation to a considerably larger salary for its members. Federalist
newspaper editor William Coleman decided to attack the bill in print on
the grounds that his readers would blame the majority party but would
never bother to inquire how individual Federalist legislators voted. On
the other side of the partisan divide, Thomas Jefferson shared the expec-
tation that individuals’ votes would not be monitored, predicting that
‘‘almost the entire mass [of Congressmen] will go out, not only those
who supported the law or voted for it, or skulked from the vote, but those
who voted against it or opposed it actively, if they took the money’’
(White 1951:401).
Both Coleman and Jefferson proved mistaken, however, as Republican
newspapers were quick to point out that a greater proportion of Federalist
than Republican members voted for the pay raise, as well as to publish the
names of the guilty parties (Skeen 1986; Bianco, Spence, and Wilkerson
1996). Public outrage fell more heavily on supporters of the act than on
52
The U.S. Experience
opponents: 19 percent of supporters were reelected against 46 percent of its

opponents (Skeen 1986). Recent research, moreover, strongly suggests that
the members of the Fourteenth Congress themselves perceived better than
Coleman or Jefferson the salience of their individual positions on the act to
voters in their districts, both before and after the vote. Legislators who had
won their previous election by smaller margins were systematically less
likely to support the act, and those who supported it were subsequently
much less likely to seek reelection (Bianco, Spence, and Wilkerson 1996).
The controversy surrounding votes on the Compensation Act included
intense newspaper coverage and public meetings in various communities.
According to Skeen (1986), the episode undermined the idea of deference
by citizens to representatives in the new republic and established the norm
of deference by legislators to public opinion instead.
3.3.3. Objections to Secret Voting

If the practice of recording and making public individual votes is as old as
the U.S. Congress, one critical nineteenth-century episode sheds light on
how public voting affects relations between legislators, party leaders, and
their constituents. At issue was the procedure for electing officers of the
House of Representatives, including the Speaker. Prior to 1839, internal
House elections were conducted by secret ballot. During the 1830s, battles
over patronage controlled by these offices incited moves by leaders of the
Democratic majority to push for public voting in House officer elections.
It is important to highlight that the impetus to record votes in this case
came from party leaders, who otherwise could not monitor the votes of
their rank and file, not from actors outside Congress. Yet, right away, the
discussion incorporated the assumptions that, if votes were recorded, they
would be made public, and that if they were made public, citizens would
take note. Fierce debates ensued between advocates of ‘‘the right of con-
stituents to know all the public acts of their representative’’ and ‘‘the dem-
ocratic principle of accountability to the constituent body’’ and defenders
of a legislator’s right to ‘‘express the convictions of his heart, separate from
party ties and party allegiance,’’ fearing ‘‘that the power of party can con-
descend to the smallest, most unimportant, and contemptible matters’’
(Jenkins and Stewart 2003:494, 495, 497).
The Democratic leaders prevailed in this initial battle, winning the ability
to monitor their members’ votes and putting a stop to the subterfuge by
majority party dissidents and cross-party coalitions that had characterized
53
many House officer elections early in the century. Yet, the effect of public
voting on party discipline, particularly for the highly salient votes to elect
House Speakers, was ‘‘exactly the opposite in the long term . . . [because] the
daylight that shone on speakership elections highlighted regional animosities
just as much as partisanship. It became more difficult to elect Speakers and
organize the House than before the onset of viva voce voting’’ (Jenkins and
Stewart 2003:504–5). The viva voce episode illustrates that, in the U.S.
context, the move away from unmonitored votes initially strengthened the
influence of national party leaders over rank-and-file legislators, but univer-
sal monitoring ultimately strengthened an even greater force, countervailing
that of party – constituents, with their diverse regional demands.
3.3.4. Interest Group Monitoring

Subsequent historical accounts demonstrate that organized interest groups
systematically monitored voting records in the early twentieth century
and that legislators feared the influence of these records on voters. In his
account of the rise of Prohibition, Peter Odegard (1928:90) quotes corre-
spondence between a state legislator and a local Anti-Saloon League’s
chapter, which sums up the politician’s simple calculus:
While I am no more of a Christian than I was last year, while I drink as much as I did
before, you have demonstrated to me that . . . there are more Anti-Saloon votes in
my district than there are saloon votes; therefore I will stand with you both with my
influence and my vote if you will give me your support.
The league, moreover, was not satisfied with fair-sounding pledges and
relied on methods of monitoring recorded votes that echo those of modern
interest groups. ‘‘Elaborate indexes of politicians and their records were
kept at Washington and in most of the states, and professions of sympathy
were matched with deeds. The voters were constantly apprised of the
doings of their representatives’’ (Odegard 1928:91). The Farm Bureau,
the American Legion, the American Medical Association, and the National
Rifle Association engaged in similar activities during this same era (Kile
1948; Mayhew 1974:66–67).
3.3.5. Electronic Voting

An important jump in the salience of recorded votes in the United States
came in 1973 with the adoption of electronic voting technology in the
54
The Supply of Recorded Votes in Latin America
House. Sponsors of the Legislative Reorganization Act of 1970 reduced the

requirements for members to demand that a vote be recorded on the
grounds that this would improve accountability of members to their con-
stituents. Shortly thereafter, to accommodate the increased demand for
recorded votes within time constraints, the House installed electronic vot-
ing machines. These changes produced a gigantic increase in the number of
recorded votes in the House, particularly on amendments to bills, and
a concurrent increase in the relative importance of voting records to legis-
lators’ relations with their constituents. One additional property of the shift
to electronic voting in the United States is worth noting. Minority-party
members – those most likely to be dissatisfied with overall legislative out-
comes – were inclined to push amendments that, when subject to recorded
votes, would be politically uncomfortable for the governing majority
(Smith 1989:29–34).
3.3.6. Lessons from the United States

Recorded voting has been integrally connected to legislative accountability
throughout the history of the U.S. Congress. Since at least the early nine-
teenth century, members expected voting records to be available and salient
to constituents, and relevant to their own electoral success. Party leaders,
too, have a keen interest in monitoring votes but, except under unusual
procedural circumstances (e.g., secret voting in House officer elections),
leaders face minimal obstacles to monitoring votes, so asymmetries in
monitoring costs generally favor party leaders over other competitors for
legislators’ loyalties. Interest groups have long treated voting records as the
currency of legislators’ performance. Finally, the reduced procedural costs
of publicizing votes that accompanied electronic voting in the House
increased their importance and amplified their relevance as a tool of the
legislative opposition.
3.4. The Supply of Recorded Votes in Latin America

In contrast to the U.S. Congress, Latin American legislatures generally
record very few votes. Beyond this straightforward observation, I
want to highlight three key points in this section. First, the supply of
visible votes in the Americas directly reflects the technological and pro-
cedural obstacles to recording and publishing votes. Second, declining
55
technological barriers have prompted procedural reforms in some

cases that facilitate the recording and publication of votes, which in
turn increases their supply. Third, this has not been so in all cases,
however; some legislatures in which the technology is available still do
not record, or else record but do not publish, which means that votes
remain invisible and effectively impossible to monitor for actors out-
side the legislature. The Appendix to this chapter recounts an episode
from my own field research that illustrates how difficult it can be to
gain access to vote records, even for a persistent and reasonably well-
connected investigator.
3.4.1. Procedures for Recording Votes

Rules of procedure in Latin American legislatures often require
recorded votes under specific circumstances – for example, on votes to
override presidential vetos in Uruguay, on the vote to select a president
in the absence of a popular-vote majority in Bolivia, and on constitu-
tional amendments in various systems – but these circumstances tend to
be unusual. In every Latin American legislature, members may also re-
quest recorded votes. The procedural barriers to such requests vary
from requiring a majority vote in Costa Rica and Bolivia, to a one-third
threshold in Peru, to petition by ten legislators in Ecuador or six in
Guatemala. Literally calling the roll in order to take votes, however, is
always a time-consuming and impractical process. Moreover, rules of
procedure for nominales often require that each legislator, in casting
a vote, be allowed floor time to justify her position. In short, logistics
alone are sufficient to rule out the traditional nominal as a means of
legislative voting throughout Latin America under all but exceptional
circumstances.
The procedures for taking standard votes (sometimes referred to as
econo´micos) vary. In chambers with smaller memberships, such as the Cen-
tral American assemblies and senates in bicameral legislatures, individuals
generally cast votes by either standing or hand raising, with a head or hand
count conducted from the mesa directiva. This procedure is impractical
when membership rises much more than 100, however, and larger cham-
bers such as Mexico’s and Venezuela’s have conventionally expedited mat-
ters by allowing party leaders to cast votes for their entire blocs. Legislators
who are present and do not explicitly state their opposition are counted as
having voted as the leadership declares.
56
3.4.2. The Frequency of Recorded Votes

Table 3.3 shows the incidence of recorded votes across twenty-four legis-
lative chambers in the Americas, plus the joint sessions of the Uruguayan
Congress. The Panamanian Assembly is included twice, under separate
voting regimes, both before the advent of electronic voting and after.
The cases are grouped according to the procedural barriers to recording –
whether an electronic voting system is used, and the threshold for requiring
a recorded vote. These two elements are connected, both logically and em-
pirically. Modern electronic voting systems automatically and instantly
generate individual-level records of votes, reducing the cost of recording,
in terms of legislative staff labor and session time, to near zero. Where the
cost of recording votes is negligible, in turn, there is less reason to maintain
rules that discourage recording.
Of twelve chambers with electronic voting systems, the rules of four
establish electronic voting as the default procedure for floor votes and
another five set request thresholds at 10 percent of members or less. Of
twelve chambers without electronic systems, only the U.S. Senate records
as standard operating procedure, two more set low barriers, and seven set
a majority request threshold to record. The mean number of recorded
votes per year is derived from collection of data from parliamentary Web
sites in those cases where complete transcripts of all legislative sessions (or,
more rarely, the votes themselves) are systematically posted and from field
research or consultation with legislative staff or academic experts in each
country otherwise.
Average number of votes per year is a fairly raw statistic, to be consumed
with some caution. The averages are derived from across thirty-three years
for Costa Rica, ten for the United States, seven for Guatemala, four-year
legislative periods for Colombia and Ecuador, and a mere nine months in
Nicaragua’s dawning electronic era. Although most of the figures are the
result of comprehensive archival searches, some of the vote totals reported
are estimates based on incomplete data (see, e.g., notes e, i, and j to Table 3.3).
Specialists in the legislative politics of each country might also reasonably
argue that legislative floor votes can have distinctive meanings in different
settings. Where most legislative work takes place in committees, for example,
floor votes may be less frequent and less central to the policymaking process
than when more of the action is on the floor. Even if we acknowledge such
qualifications, however, these votes are critical to legislative decision making
in every chamber. Floor votes are where statutes, budgets, treaties, veto
57
Table 3.3. Effects of Procedure on the Availability of Public Voting Records
Country Chamber Members Request Threshold Votesa

Standard operating procedure, electric
Chile House 120 Art. 9 328
Chile Senate 47 Art. 157 45
Nicaragua Unicameral 90 Rules do not yet reflect 924b
adoption of electronic
voting (Arts. 104–10)
Peru Unicameral 120 Art. 57 540
Standard operating procedure, manual

United States Senate 100 Rule XII 350
By request, electric
Mexico House 500 6 legislators (Art. 148) 155
Mexico Senate 128 6 senators (Art. 148) 156c
Panama Unicameral 71 Majority of those present 144d
(2004) (Art. 196)
Brazil House 513 6%, or party leaders >68e
representing 6%
of members (Art. 185)
Argentina House 257 10% of deputies present 17f
(Art. 190)
United States House 434 20% of quorum (Rule XX) 559
Brazil Senate 81 Majority of those present 125g
(Art. 294)
Venezuela Unicameral 165 Majority of those present 0h
(Arts. 120, 125)
By request, manual
Guatemala Unicameral 140 6 legislators (Art. 95) 8.4
Ecuador Unicameral 100 10% of legislators (Art. 70) 4.5
Bolivia House 130 Majority of those voting 0
(Art. 107)
Bolivia Senate 27 Majority of those voting 0
(Art. 116)
Uruguay House 99 One-third of those present <1.0i
(Art. 93)
Argentina Senate 72 Majority of those present 21j
(Art. 205)
Colombia House 161 Majority of those present 2.5
(Art. 146)
Colombia Senate 102 Majority of those present 2.0
(Art. 146)
58
Country Chamber Members Request Threshold Votesa

Costa Rica Unicameral 56 Majority of those present 0.5
(Art. 101)
El Salvador Unicameral 84 Majority of those present 0
(Art. 37)
Panama Unicameral 71 Majority of those present 1.9k
(1991–2003) (Art. 196)
Uruguay Senate 31 Rules allow, but do not <1.0i
specify procedure to
request, recorded
vote (Art. 100)
Constitutional requirement, manual
Uruguay Joint session 130 Recorded vote required on 6.3i
motion to override
presidential veto (Art. 141)
a
Recorded-votes-per-year averages are based on the time periods listed in Table 5.3, which describes data included in
the analysis of voting unity in Chapters 5 and 6. Cases without recorded votes (i.e., votes/year ¼ 0) are based on the
author’s observation at the time field research was conducted, during the early years of the 2000 decade.
b
An electronic voting system installed in 2000 was immediately put into regular use for all votes. The vote records,
however, are not published.
c
The Mexican Senate Web site currently publishes all recorded votes, but leaves only those from the most recent
session on the Web site, continually replacing records of previous votes. As a result, only a handful of votes are
available at any given time, and any actor who seeks to monitor Senate voting must be vigilant and assiduous in
harvesting votes as soon as they are ‘‘ripe.’’
d
As of September 2004, the National Assembly began to use its electronic voting system and to post electronic votes on
its Web site (http://www.asamblea.gob.pa/portada.asp). During the last three months of 2004, 36 votes were
recorded, a rate of 144 per year.
e
Actual number is somewhat higher. Figueiredo and Limongi’s (2000) data included 678 votes over a ten-year period,
but excludes votes in which less than 10 percent of deputies voted on the losing side.
f
Source: Asociación pro los Derechos Civiles, Buenos Aires.
g
Source: Scott Desposato, personal communication.
h
Art. 125 states that a single deputy may solicit a recorded vote, but does not stipulate a requirement for the approval of
such a request. Art. 120 states that Assembly decisions are to be made by vote of a majority of those present. The
transcripts of floor debates available in the Diarios de Debates online report only aggregate vote totals, not individual-level
voting records, even for votes said to be taken by the nominal method. Thus, no records are public.
i
Extensive search of Diarios de Sesiones, 1985–94, turned up only a handful of recorded votes in either chamber, beyond
63 from joint sessions on votes to override presidential vetoes, per constitutional requirement. Method of archiving
makes it difficult to determine whether any other recorded votes exist. Sources: Scott Morgenstern, Daniel Buquet,
and Daniel Chasquetti.
j
The Asociación por los Derechos Civiles reports that the Argentine Chamber produced 17 recorded votes in 2003, and
Clarıń (2004a) reports that the Chamber recorded votes in a total of nine sessions. The same article reports that the
Argentine Senate recorded votes in a total of eleven sessions. The estimate of 21 votes is based on an assumption that
the Senate recorded votes with the same frequency per session, given that records were produced.
k
Source: Harry Brown Araúz, personal communication.
59
overrides, and constitutional amendments are ultimately approved or

rejected, and the availability of vote records indicates how much hard infor-
mation citizens have about the most consequential actions of their represen-
tatives.
The connection between the procedural obstacles to recording votes
and the amount of such information available is not surprising, but it is
striking nonetheless. Those chambers where electronic voting is standard
operating procedure average 459 recorded votes per year; the U.S. Senate
(standard operating procedure, manual) averages 350; chambers with elec-
tronic systems but where recorded votes must be requested by some subset
of legislators average 153 votes; and those where voting is manual and
recorded votes must be requested average about 2.
3.4.3. From Electronic Voting to Visible Voting

The experiences of individual countries that have adopted electronic voting
suggest that, once systems are in place, demands grow to alter rules of
procedure to facilitate recorded voting, and that where these demands
are successful, the numbers of recorded votes will rise, and pressure to
make votes visible increases. Electronic voting systems were adopted in
the lower legislative chambers more or less concurrently with the return
to democracy in the mid-1980s in Argentina and Brazil and in the 1990s
in Chile. They have been in use in Mexico and Peru since 1998, Nicar-
agua beginning in 2000, and in Panama since late 2004. A system has been
in place in Venezuela since 1997 but has not yet been used. The same is
true for Costa Rica’s first-generation system, installed in the mid-1970s,
and for the systems in both chambers of the Colombian Congress, in
place since the late 1990s. See Table 3.4.
There are purely pragmatic reasons to adopt electronic voting. It yields
a faster, more accurate count than hand raising. It is a concrete manifesta-
tion of modernization, an ideal widely embraced in the abstract by govern-
ments and legislative leaders (Cevasco Piedra 2000). The impact of
electronic voting, however, is potentially more substantial than the logistics
alone imply. Electronic voting generates records of individual legislators’
votes. If the records are available to the public, this effectively transforms
all votes into nominales – matters of record that individual legislators could
be called upon to defend.
There is no guarantee that journalists, interest group leaders, activists,
and the like will register how legislators vote or that constituents will pay
60
Table 3.4. Electronic Voting Systems in Latin American Legislatures

(lower chambers or unicameral systems)
Country Installed In Use? Availability of Electronic

Vote Records
Argentina Prior to redemocratization, Yes Retained in congressional
1983a archives, not published
Brazil Prior to constitutional Yes Camara dos Deputados
assembly,1987 b Plenarioc
Chile With construction Yes In text
of new Congress, 1990 of Boletin de Sesiones d
Colombia 1999 No NA
Costa Rica Mid-1970se No NA
Mexico 1998 Yes Gaceta Parlamentaria f
Nicaragua 2000 Yes Retained by Assembly
clerk, not published
Panama Late 1990sg No Asamblea Nacional Panama h
Peru 1998 Yes Congreso de la Republicai
Venezuela 1997 No NA
a
Personal communication with Mark P. Jones (Michigan State University).
b
Personal communication with Octavio Amorim Neto (Instituto Universitário de Pesquisas
do Rio de Janeiro).
c
http://www.camara.gov.br/Indice.asp?Endereco¼Intranet/Plenario/Plenario.htm.
d
http://www.camara.cl/, on line since 1996.
e
Interview with Fernando Castillo.
f
http://gaceta.cddhcu.gob.mx/.
g
Personal communication with Harry Brown Araúz.
h
http://www.asamblea.gob.pa/portada.asp, on line since 2004.
i
http://www.congreso.gob.pe/index.htm.
attention. Without a record, however, the prospect is moot, and until

recently there was no record most of the time. The adoption of electronic
voting means that records are being created in many places, and the
existence of records opens the possibility that the information will enter
into political discourse.
What rough data are available suggest a positive relationship between
visible votes and legislative individualism. The surveys of legislators in
Latin America discussed in Chapter 2 included a question on whether party
leaders should always impose discipline on rank-and-file legislators or
never impose discipline (i.e., the discretion on how to vote should always
remain with the individual), or whether discipline should depend on what
issue is at hand (Proyecto Élites Parlamentarias Latinoamericanas 2006).
61
40
Bolivia
Chile
Costa Rica
Mexico Peru
El Salvador
20
Venezuela
Individualism
Argentina
Uruguay
0
Guatemala
Colombia
-20
Nicaragua
Ecuador
-40
0 5 10 15 20 25
sqrtVV
Figure 3.1 Recorded votes (square root) and preferences for legislative
individualism.
I constructed an index of overall preference for individualism in each coun-

try by subtracting the percentage preferring ‘‘Always discipline’’ from that
preferring ‘‘Always individual discretion’’ (assuming the ‘‘Depends’’ res-
ponse to be neutral). Figure 3.1 plots this index against the square root of
the average number of votes recorded and made public each year (based on
the data in Table 3.3) for those lower chambers for which both recorded
vote and survey data were available.
The figure shows an overall positive correlation (.29) between
recorded votes and legislators’ expressed preferences for individualism.
This result is consistent with the cross-national pattern by which legis-
latures with plentiful recorded (and, generally, visible) votes are also those
in which legislators express more support for individual autonomy from
party leaders. The correlation by itself does not demonstrate which way
causality runs between visible voting and individualism – that is, whether
individualism (whatever the source) encourages the adoption of visible
voting procedures, or visible votes (adopted for whatever reason) increase
individualism – or whether the phenomena are mutually reinforcing. The
next sections describe the link between the adoption of electronic voting
and visible voting in two legislatures where the move was contentious, and
62
the next chapter explores the sources of support for, and opposition to,
visible voting more generally.
3.4.3.1. Peru. The prospect that visible voting would increase the
premium on individual legislator accountability was explicitly on the
minds of Peruvian legislators on both sides of the reform debate in 1998,
as they considered the implications of switching from the traditional hand-
raising method of voting to the electronic system, which had recently been
installed as part of a broader government modernization plan. On September
24, in an effort to embarrass the pro-Fujimori majority on a motion related to
a corruption investigation, the opposition demanded that the electronic
voting machines be incorporated into standard legislative procedure:
The whole reason for electronic voting is so citizens know how their representatives
voted, so [votes] can be publicly justified. It’s an instrument of democracy and trans-
parency, which is why Congress spent as much as it did [to have it installed], not so we
can use it on some votes and not on others. . . . What the country is going to notice is
that the parliamentary majority is afraid that, through the Internet and other mech-
anisms, its votes on some matters will be made visible. (Congreso del Perú 1998)
The opposition threatened procedural maneuvers designed to grind

progress on all matters to a halt if the electronic system were not employed.
The majority eventually broke ranks, with one of its members concurring
on the matter of transparency and accountability: ‘‘One reason for this
system is that it leaves a record of votes for current political analysts and
for history, so that how each one of us voted is known; and those congress-
persons that run for reelection, when they face the voters, they’ll have to
explain how it is that on each of the issues they voted as they did’’ (Con-
greso del Perú 1998). Soon after the old system was breached, the Peruvian
Congress began posting records of all electronic votes on its Web site, at
a rate of more than 500 per year.
3.4.3.2. Panama. A similar dynamic appears to have played out in the

Panamanian National Assembly in 2004. Early that year, during the
waning months of a legislative term, opposition deputies won initial
committee approval for a reform of chamber procedures to facilitate
use of electronic voting and to eliminate altogether the practice of
voting by legislators’ banging on their desks (loudest group wins).
Supporters of the proposal outside the assembly echoed the pro-
transparency arguments of Peruvian opposition legislators that signal
63
voting diluted accountability and allowed legislators to dissemble about

their positions on unpopular initiatives. As in Peru, the leadership of the
majority coalition initially resisted using the electronic system (Tapia G.
2004). By September of that year, however, after an intervening election,
the new National Assembly, now controlled by a former opposition party
that had campaigned on an anticorruption, pro-transparency platform,
had adopted electronic voting and begun to post votes on its official
Web site.
Panama is the only case for which I have systematic data for the
rate of recorded votes both before and after adoption of electronic voting,
and the difference is dramatic. In the seven years from 1991 to 1997,
without electronic voting, the Assembly recorded 11 votes, at a rate of
1.6 per year. During the last three months of 2004, it recorded 36 votes,
an annual rate of 144.
3.4.4. Why More Votes Are Not Recorded, and Why Recorded Votes
Are Not Always Visible
Even if we take technological and procedural barriers into account, the
Latin American legislatures generally record fewer votes than the U.S.
Congress. After all, the U.S. Senate has no electronic voting, and House
rules impose a significant request threshold, yet both record hundreds of
votes per year. One obvious explanation is a lack of staff resources in Latin
American legislatures without electronic systems. Legislatures in the
region are chronically and notoriously underfunded, and recording votes
by hand is labor-intensive and time-consuming.
Procedural costs cannot be the whole story, however. Even where elec-
tronic systems are in place in Latin American legislatures, their use is not
a given. Electronic systems are in place in the Costa Rican and Venezuelan
assemblies, but they are never used, just as in Panama until 2004, and the
electronic systems in the Argentine and Colombian lower chambers are
very rarely employed. In other cases, the systems are used regularly, but
voting records are not systematically published. The Nicaraguan Assem-
bly, for example, records all votes, but does not publish the records. The
Argentine Chamber’s Web site puts up the aggregate (yes, no, abstain,
absent) results for votes taken electronically but includes the lists of in-
dividual deputies’ voting decisions only sporadically. The Mexican Senate’s
Web site, curiously, publishes its recorded votes the evening of each given
session but then removes the records when the votes from the next session
64
Conclusion
are put on line. The Mexican Chamber’s Web site has changed its policies
for making recorded voting data available various times during the early
years of the 2000 decade before settling on putting up complete records of
electronic votes.3
Public access to information about electronically recorded votes may be
partly attributable to technical capacity. Maintaining a comprehensive
Web site taxes the resources of many assemblies. On the other hand, party
leaders and the members of dominant coalitions sometimes prefer not to
make voting records public even when they are kept and not to use elec-
tronic voting systems even when they are in place. For now, it is worth
noting that, although the point of departure is different, the relationship
between electronic systems and recorded voting in Latin America runs in
the same direction as that in the United States, albeit starting from a dif-
ferent point. In the United States recorded voting is common even in the
absence of electronic voting, and it became more prolific with the adoption
of an electronic system in the House of Representatives. In Latin America,
recorded voting is negligible in the absence of electronic voting. It is in-
creasingly common – but not a given – where electronic systems reduce
procedural costs.
The experiences of individual countries that have adopted electronic
voting systems affirm the pattern evident in the cross-national data: elec-
tronic voting and minimizing procedural barriers to recorded voting boost
the amount of information available to those outside the legislature about
legislative decision making.
3.5. Conclusion
Legislative voting can be an element of accountability only if votes are
publicly known. They can be known, in turn, only if they are recorded
and the records are available. Voting records are so essential to how legis-
lative accountability is conceived in the United States, and have been for so
long, that it is easy to forget that there is nothing automatic about their
relevance, or even their existence and availability. Beyond the United
States, essential components of the legislative voting–legislative account-
ability relationship are often missing.
3
The practices described in this paragraph refer either to convention at the time field re-
search was conducted for this project, during the first years of the 2000 decade or, in the
references to Web site content, to practice up through the middle years of the decade.
65
In the Latin American context, systematic voting records never exist in

the absence of electronic voting. Given this empirical fact, it is important
that reliable electronic voting equipment be more easily accessible than
ever before. The technical obstacles to keeping and maintaining legislative
voting records are rapidly diminishing. This, in turn, makes it increasingly
feasible for advocates of visible voting to push their case, and difficult for
opponents to resist demands to record and publicize legislative votes.
Diminishing technological barriers do not, however, mean recorded voting
and public voting have been, or will be, welcomed into all legislatures as
standard operating procedure, as the next chapter makes clear.
Chapter 3 Appendix: Stalking the Elusive Recorded Vote

During field research in Bogota, Colombia, in May 2001, various legisla-
tors assured me that the office of the secretary-general of the House of
Representatives maintained computerized records of all electronic votes
from the previous five years or so and that these could be made public.
The aide to one of these legislators managed to deliver me all the way to the
secretary-general himself, who listened uneasily to my request, but assured
me that all information is public, and handed me off to his technical assis-
tant, who, as I initially understood it, would produce for me computerized
files of the votes.
The technical assistant took me to the House chamber itself to show me
the voting machines, which are used regularly to take attendance but less
frequently to record votes. The machines are modern and recognize the
representative’s hand, rather than fingerprint. (The assistant noted that the
larger hand recognition slots are more trouble than the smaller fingerprint
recognition type because when legislators eat lunch at their desks, they tend
to get pieces of food stuck in there, which are difficult to get out. He looks
forward to the day they can be replaced with a fingerprint recognition
system.) The assistant also told me that the electronic system had originally
been installed in the House chamber in the 1980s but was modernized in the
late 1990s. He said that the Senate had installed the exact same system shortly
after the House, but that the senators never use it. He did not know why.
Gradually, it became clear to me that the technical assistant had not been
instructed that his task was to provide me with computerized files of votes.
When I asked him directly, he told me he did not think the office of the
secretary-general even had complete records of votes, in part because some
of the more Luddite legislators did not like to use the electronic machines,
66
Appendix
even when nominales are taken, and instead write down their votes and pass
them forward. The complete record of any nominal, therefore, is actually
assembled by the chamber’s Division of Recording and Dissemination.
Having established exactly what I was seeking, the technical assistant to
the secretary-general took me to the director of the Division of Recording
and Dissemination. This director listened to my request for electronic
records of votes, and then sent me, with his assistant, to the deputy secretary
of the House chamber, who, I was assured, was the guy who knows where
the computer files containing votes are, and how they are organized.
The deputy secretary listened to my request and told me that records of
all nominales are published in the Gaceta (the official published record of
legislative proceedings), which I already knew. I explained that I hoped to
avoid having to go through every page of every Gaceta to find the rare
nominales, make photocopies, and then transcribe each vote to a spread-
sheet. I told him that I had assumed electronic records must exist – perhaps
even of more votes than the formal nominales – given the existence of an
electronic voting system. I noted that even the records published in the
Gaceta are written up on computers prior to printing, that electronic copies
would be far more convenient than print versions. And I pointed out that
various sources – legislators and staff – within the House of Representatives
itself had confirmed the public availability of these resources. The deputy
secretary countered that the electronic voting system does not produce
a record of the votes; then clarified that the system does produce a record,
but that the record is deleted from the system immediately after each vote is
taken and the result of it is printed up. Subsequently, he explained, the
Division of Recording and Dissemination transcribes from these printouts
a new list of who voted how, which is inserted in the Gaceta. The sub-
secretary explicitly did not offer to make electronic records of the Gaceta
available. Rather, he reiterated his offer to let me – or anyone, because this
information is public, after all – leaf through the hard copies page by page
to find the votes and make photocopies.
After leaving the deputy secretary’s office, the assistant from the
Division of Recording and Dissemination (still with me) told me that the
director of the Legislative Archive, or possibly director of the Gaceta, would
likely be more cooperative, and that I should check with them. I took this
advice, but with results that should, by now, be predictable. Pursuit of legis-
lative voting records in many other countries, including Bolivia, Ecuador,
and El Salvador, yielded similar results.
67
4
Demand for Visible Votes
4.1. Is Transparency Desirable?

Democratic legislatures are forums for debate and collective decision over
the diverse values their members represent, and their internal workings are
generally meant to be subject to monitoring from outside actors. By forcing
debate into an open setting, legislatures may limit admissible arguments on
behalf of interests or policy positions to those that can be defended in
public. Transparency in voting, moreover, opens the possibility that in-
dividual representatives can be held accountable for their votes by those
they represent.
The argument for transparency in legislative voting rests on twin ideas.
One is that political elites and ordinary citizens differ in their claims to
anonymity in political action. The other is that information about legisla-
tive voting actually gets to voters. According to the conventional logic,
anonymity is necessary for voters, through the secret ballot, to free them
from intimidation in elections, but in legislative voting it undermines dem-
ocratic accountability. In effect, legislators ought to be subject to pressure
on their votes, but citizens should not (United States Supreme Court
1958).1
The distinction between legislative voting as public and elections as
private is not universally shared. Jean Jacques Rousseau, for example, did
not advocate public voting in legislatures but nevertheless makes a case for
public voting in elections. Rousseau regarded individual legislative votes as
of less interest than the collective result. Whether the latter was consonant
1
For a contrary argument, that even citizen voting at the polls should be public, see Brennan
and Pettit (1990).
68
Is Transparency Desirable?
with the general will, according to Rousseau, could be determined only

through popular election, and in this forum Rousseau prioritized the ac-
countability of citizens to each other, holding that the requirement to state
one’s vote publicly could encourage citizens to support only proposals that
are in the general interest (Rousseau 1763).
Neither is the principle that information regarding legislative votes
ought to be available to voters a given. As Chapter 3 demonstrates, empir-
ically the supply of voting information is spotty at best in many legislatures.
Moreover, at around the same time as Rousseau, but drawing on an alto-
gether different conception of representation, Edmund Burke rejected the
idea that legislative representation ought to consist of legislators’ faithfully
reflecting their constituents’ preferences or answering to voters for failing
to do so. Burkean representation, as famously depicted in his Speech to the
Electors of Bristol (1774), has no place for legislative transparency and the
monitoring of votes by those outside parliament. Burke pointed out to his
constituents the narrowness of their own political vision and demanded
that they trust their representative’s (i.e., his) judgment as to the common
good and forgo demands for responsiveness to their own wishes.
In short, political theory can supply normative arguments against leg-
islative transparency, but this book subscribes to the belief, mainstream to
much modern political theory, that transparency makes accountability pos-
sible and that accountability of representatives to citizens is desirable.2 All
the evidence in this chapter suggests that the normative commitment to
transparency is widely shared, or at least outwardly acknowledged, among
legislative insiders and outsiders alike. Advocates of visible voting regularly
echo this normative argument, and even those who are reluctant about
visible voting do not directly challenge it. The main obstacles to visible
voting in modern legislatures are not normative arguments but other fac-
tors. In order to understand the empirical pattern of recorded and signal
voting described in Chapter 3, it is necessary to examine the political forces
and preferences on the matter at play in legislative politics.
2
For example, Snyder and Ting (2005) offer a formal model to explain why public voting in
legislatures ought to be appealing both to legislators and to citizens. The outcome depends
on symmetrical monitoring of legislators’ voting behavior by both their constituents and
some other principal – say, a powerful interest group or a political party leader – whose
preferences may be at odds with constituents’. Symmetrical monitoring is necessary for
legislators to make binding electoral contracts with their constituents – to commit credibly
to be faithful representatives and to expect any reward for doing so. A critical assumption of
the model is that information about legislators’ votes gets to constituents.
69
This chapter advances a series of propositions regarding differences across

political systems that shape the electoral salience of voting records and differ-
ences among actors within political systems in their preferences for making
votes public. Then I examine these propositions in light of evidence from field
research, interviews with legislators, and archival and secondary sources.
4.2. Incentives to Monitor and Publicize Votes

I distinguish between two types of factors that affect the impetus toward
visible voting. The first operates at the level of the political system –
specifically, the manner in which legislators are elected – and shapes the
incentives legislative candidates have to emphasize voting records in cam-
paigns. The second distinguishes among political actors within a given
political system according to their inherent ability to monitor votes even
in lieu of recording and publication (i.e., signal votes, which are the pro-
cedural default in most legislatures), as well as their relative inclination to
answer to principals whether inside or outside the legislature.
4.2.1. SMD versus MMD Elections

Citizens cannot be expected to keep track of voting records on their own,
even where votes are published regularly in places easily accessible to citi-
zens. Common sense dictates that all but the most peculiar have better
things to do than to comb through records of legislative proceedings on
a regular basis.3 Without relying on the initiative of citizens themselves,
then, how might information about the quality of representation delivered
via legislative votes get into the hands of voters so that it can contribute to
accountability? In the U.S. context, it is clear that politicians are motivated
to deliver information about legislative voting records to voters, but this is
less obvious for many legislatures elsewhere.
3
Although Burke would eventually prove to be among the most sympathetic members of the
British Parliament to American grievances against the crown, many of the colonists did not
share his views about deference to legislators. One particularly un-Burkean reform ad-
vanced by reformers in Pennsylvania around the eve of independence would have required
any proposed law to be posted in public houses for a year before the colonial legislature
could vote on it, to ensure citizens could monitor their representatives in advance of acting
(Morone 1990:40). The Internet notwithstanding, these Founders appear to have had
a keener sense than we do at present of how to use available technology to foster account-
ability.
70
Incentives to Monitor and Publicize Votes
Candidates in single-member-district (SMD) elections dominated by

two parties have strong incentives to provide voters with bad news about
their opponents’ records because any candidate is the primary residual
claimant of whatever popular vote loss her opponent suffers as a result.
In the context of U.S. elections, challenging candidates are key vehicles by
which information on legislative voting records are delivered to voters in
a legislator’s district.
In multimember-district (MMD) electoral systems, which predominate
in Latin America, the incentives for candidates to deliver news about their
opponents’ voting records are weaker. In closed-list systems, such as those
of Argentina, Costa Rica, Ecuador, El Salvador, and Nicaragua, ballots do
not afford voters the opportunity to indicate a preference for individual
legislators, so any punishment for bad voting behavior (or reward for good
behavior) is spread across all of a party’s candidates, and any benefit to the
lists of other parties accruing from such punishment may be spread across
the other parties’ running lists. Within each of these lists, any gain in
electoral support benefits the candidate highest in the list who would other-
wise not be elected. The identity of this marginal candidate, however, is
unlikely to be known with certainty before the election.
Systems that allow for personal preference voting over candidates within
parties, such as the open lists in Brazil, Peru, Chile, and Colombia,4 provide
stronger incentives for the delivery of information about voting records but
still not as strong as under SMD because of the diffusion across multiple
candidates of any gains to be had from exposing flaws in a given legislator’s
record. In the concrete sense, a candidate in a Peruvian district with eight
incumbent legislators has multiple voting records to criticize but cannot
expect to monopolize whatever electoral support she dislodges from
incumbents by publicizing their negligent voting behavior. Instead, this
support may be spread across the other challengers for the eight seats at
stake.
In short, even in personal vote systems, the potential gains to be had
from publicizing ‘‘bad’’ voting by an opponent depends on how widely any
votes shaken loose by such publicity will be dispersed among other candi-
dates. How many candidates there are in the district will likely depend, in
turn, on the district magnitude, the fragmentation of the party system, and
how may candidates each party nominates. By and large, the incentives of
4
Open lists were used in Colombia beginning with the 2006 election, but the previous system
of multiple lists was at least equally personalistic (Cox and Shugart 1995).
71
campaigns in large-magnitude districts are to undersupply information

about individual legislative behavior (Desposato 2004a, 2006a). Generally
speaking, the smaller the pool of candidates chasing votes in a district, the
greater the incentive for candidates to emphasize their opponents’ individ-
ual voting records in campaigns.
The incentives for candidates to make voting records central to cam-
paigns are strongest under the electoral rules and party structure that char-
acterize U.S. legislative elections. Purely electoral incentives for candidates
to publicize their opponents’ voting records are not altogether absent in
other electoral environments – and citizens may well come to monitor
legislators’ votes for other reasons – but the structure of electoral compe-
tition in MMD systems provides less incentive than in SMD systems to
make voting records central to political debate. Legislative voting proce-
dures may determine who has the ability to monitor legislators’ actions, but
electoral rules and the nature of party competition shape the incentives for
politicians to deliver this information.
4.2.2. Legislative Insiders versus Outsiders

Recording and publishing votes provide less marginal benefit to actors who
enjoy natural advantages in monitoring signal votes (call these ‘‘insiders’’)
than to those who are relatively disadvantaged (‘‘outsiders’’). To the extent
that insiders and outsiders compete with each other for influence over
legislators, then insiders should oppose making votes visible while out-
siders should support visible votes. With respect to legislators themselves,
they know that insiders will be able to monitor their behavior, rewarding
and punishing accordingly; the question at stake in recording and making
votes public is whether outsiders will be able to do so as well. Thus, those
who seek to appeal to outside audiences and constituencies – and particu-
larly those inclined to resist coercion from insiders – should support visible
voting, whereas legislators whose foremost principals are insiders should
oppose visible voting, or at least be indifferent (Snyder and Ting 2005).
The political actors with the greatest monitoring advantages are legis-
lative party leaders – particularly those from majority parties or coalitions
that control legislative agendas. Interest groups with substantial resources,
including lobbyists or staff poised to oversee legislative activity, may also
be able to monitor signal votes on the assembly floor. Actors motivated
to monitor, but with disadvantages in doing so, include nongovernmental
and watchdog organizations with constrained resources, journalists, and
72
Incentives to Monitor and Publicize Votes
academics. The actors most resistant to coercion from legislative leaders

are opposition legislators and dissidents within majority parties and coal-
itions. Interviews and documentary evidence confirm that the politics
of recording votes and making records transparent conforms to these
expectations.
4.2.3. Friends and Enemies of Visible Voting

Three propositions follow from the discussion of visible voting in this
section. First, the purely electoral incentive to emphasize an opponent’s
legislative voting record should be stronger in SMD than in MMD systems,
contributing to somewhat greater salience of visible votes in the former
than in the latter. Second, various sorts of political actors may be inclined to
monitor legislative votes – for example, constituents, political party leaders,
and interest groups. Among these actors, those inside the legislature (e.g.,
party leaders) for whom monitoring votes is relatively cost-efficient have
little interest in formally recording votes relative to those outside the leg-
islature, for whom monitoring costs are higher (e.g., constituents, interest
groups). Third, among legislators themselves, visible voting should be
favored by those inclined to resist pressure from inside actors and opposed
by those in a position to apply it.
In light of the last proposition, consider the relative attraction of visible
voting and the common alternative, signal voting, for legislative party lead-
ers and for rank-and-file legislators, respectively. A variety of factors might
come into play here, including the mundane procedural costs of recording
a vote, the value of signaling loyalty to (or independence from) one’s party,
and the value of being on the winning side. In traditional, low-tech legis-
latures, signal voting imposes lower procedural costs than does recording
and publishing individual votes. These time and labor savings are salient
especially to party leaders, who generally manage the administration of
their chambers. However, technological changes that make electronic vot-
ing technology cheaper and easier are erasing the procedural advantage of
signal over visible voting.
Next, consider the signaling value of individual votes among insiders.
Within the legislature itself, signal and visible votes are equally visible to
other legislative actors. Leaders can monitor, and rank-and-file members
can express fealty, or dissent, to their leaders under either format. The
situation is different with respect to signaling outsiders, however. For lead-
ers who bear primary responsibility for maintaining their party’s collective
73
reputation, responsiveness to outside actors whose demands might conflict

with the party line is a liability. Signal voting protects legislators from such
demands, so it is preferable to visible voting. For rank-and-file legislators,
the situation is less clear-cut. The anonymity of signal voting, and the
protection it affords from pressure by legislative outsiders, may be attrac-
tive in some instances. But so may be the flexibility visible voting allows to
signal individual vote positions (even dissent from the party line) to an array
of potential principals, including voters, interest groups, governors, or
presidents.
Finally, particular policy questions aside, consider the value of winning
itself. The positions of parties, and their legislative leaders, are generally
known to journalists and disseminated in the media, so whether a party is
seen to be on the winning side does not depend on whether a vote is visible
or not. However, if a party loses because its members split on a given vote,
visible voting will expose the split to audiences outside the legislature,
compromising leaders’ reputations for authority and competence, a matter
of less concern to rank-and-file legislators. The consistent pattern here is
that visible voting presents a potential liability to party leaders – the con-
summate legislative insiders – whereas its costs to rank-and-file legislators
are less and its appeal is considerable.
4.3. How the Political Actors See Things

The discussion in Chapter 3 demonstrates that visible legislative votes have
long been ubiquitous, and politically important, in the United States. The
availability of individual voting records far surpasses that of any Latin
American legislature. This situation, together with the demonstrated cen-
trality of legislative votes to electoral accountability, supports the propo-
sition that visible votes are more salient in the United States, where
elections are exclusively SMD, than in Latin America, where MMDs pre-
vail. In this section, I consider the varying demand for visible votes among
actors in legislative politics within Latin American systems, where field
research for this project was conducted. This research confirms the intu-
ition that insiders, for whom the opportunity costs of monitoring are low to
begin with, tend to oppose voting transparency, whereas outsiders and
legislative dissidents support it. Legislative party leaders – the ultimate
insiders – are reluctant to use institutional resources to eliminate the very
monitoring asymmetries that signal voting implies. Actors outside legisla-
tures, for whom costs of monitoring are otherwise high, favor public
74
How the Political Actors See Things
voting. Among legislators, members of majority parties or coalitions – and

most vehemently their leaders – tend to oppose public voting, whereas
opposition legislators and dissidents within majority coalitions support it.
4.3.1. Legislative Leaders

In enforcing party discipline, legislative party leaders have an inherent in-
terest in monitoring votes.5 However, they are best off if they can monitor
effectively without formally recording, insofar as the absence of an official
record that can be examined by outside actors shields their rank-and-file
legislators from competing pressures. Party leaders’ interest in recording
votes, if it were to exist at all, would stem from their inability to keep track
of how their groups vote by other means (Cárdenas interview; Jenkins and
Stewart 2003).
Most legislators interviewed suggested that formally recording individ-
ual votes is not necessary for leaders to monitor their troops, and all the
party leaders interviewed found informal methods of monitoring votes to
be sufficient for their needs, consistent with the proposition that the mon-
itoring advantages are a valuable resource for legislative insiders under
signal voting. Former Salvadoran Assembly president Juan Duch (ARENA)
explained that ‘‘we don’t count with an electronic system, but it is still easy
to know how a party group voted, and therefore one can know with near
certainty whether there is a majority, and whether it is a simple or extraor-
dinary majority.’’
Other party leaders pointed to the ergonomics of how seating and voting
in their chambers operate. Carlos Vallejo Lopez, former president of the
Ecuadorian Congress, for example, noted that, ‘‘Because the party group is
almost all in a line – they are physically together – one can observe how the
bloc moves. Ten deputies raise their hands, five deputies, whatever.’’ Legis-
lators from that and other chambers – party leaders and back-benchers,
governing coalition and opposition – made similar points about party
leaders’ ability to monitor voting (Lucero, Sibaja, and Vargas Pagán
interviews).
It may be that such low-tech monitoring is more feasible in smaller than
in larger legislatures. Most Latin American legislative chambers have fewer
than 150 members. The lower houses of Argentina, Brazil, and Mexico, on
5
Legislative parties in Latin America are referred to variously as bloques, fracciones, bancadas,
grupos, or partidos. For simplicity, I refer to such units generically as party groups.
75
the other hand, have 250 to more than 500 members, and all three have
electronic voting systems installed. Chamber membership is positively
correlated (.45) with use of electronic voting, but the correlation between
membership and frequency of recorded votes is much smaller (.15), and not
significant. Party leaders in all chambers where interviews were conducted
emphasized that discipline is expected in voting, so recording individual
legislators’ votes would be redundant (R. Alvarenga and Duch interviews).
According to Gonzalo Sánchez de Lozada, former president of Bolivia,
‘‘Here it’s not like the United States where you say ‘I voted this way or
that.’ . . . Here, people presuppose and expect legislators to be loyal to
their party.’’
Overall, party leaders interviewed were consistently dismissive of the
need to record and publish legislative votes. Press reports from other legis-
latures confirm this pattern. The president of the Panamanian National
Assembly dismissed a proposal by opposition deputies to adopt electronic
voting on the grounds of unspecified technical problems with the system
and the lack of training in its use among the deputies themselves (Tapia G.
2004). After an intervening election, however, the next Panamanian As-
sembly saw fit to begin recording and publishing votes (Asamblea Nacional
de Panamá 2005). Similarly, in 2003 party leaders in the Texas House of
Representatives initially resisted demands to record and publish votes on
the grounds that to do so would be too costly in terms of time, despite the
fact that an electronic voting system was already in place, and in terms of
the extra paper required to include the voting rolls in the published House
Journal (Dallas Morning News 2003a). Neither factor proved pivotal in the
longer run, however; by 2005 Texas votes were available on the state leg-
islative Web site (Texas Legislature Online 2005). In contrast to leaders,
legislators further removed from positions of control over the agenda ex-
press consistent support for recorded voting. Many of these also suggested
that party leaders are actively hostile, rather than indifferent, toward
recorded voting.
4.3.2. Opposition and Dissident Legislators

Legislators from opposition parties, as well as occasional dissident legis-
lators from within governing parties or coalitions, are dependable support-
ers of the idea of recorded voting, whether or not recorded voting was the
standard procedure in their chamber. Their motivations fall into four cat-
egories: to obstruct the legislative process, to prevent the manipulation of
76
voting results, to hold their adversaries’ feet to the fire on votes that are
unpopular, or to publicize dissidence within the majority party or coalition.
4.3.2.1. Obstructionism. The first motivation, obstruction, is relevant

only in chambers that lack electronic voting systems. Where recording a vote
literally means that the roll must be called – and, in some cases, where legis-
lators are allowed to explain their position, however briefly – the effect can be
to grind progress on the legislative agenda to a halt (Cárdenas interview).
4.3.2.2. Honest Tallies. The second motivation is simply that recorded

votes prevent outright manipulation of voting results by those who
control legislative procedures. One Ecuadorian legislator, who asked for
anonymity with regard to this one comment, said:
The truth is the following: the president of the Congress often manipulates the
votes. So, when you don’t have a nominativa . . . if the secretary says there are sixty-
four votes out of the 110 delegates who are present in the hall, the article is passed.
[My party] usually has two of its assistants in the Congress at the front, on both sides
of the plenary hall. . . . It’s a warning for the secretary, because on more than one
occasion we’ve demonstrated with the accounting that we have brought with us that
they are giving a result that’s incorrect. [See also Lucero interview.]
Salvadoran deputy Aristides Alvarenga (Partido Democrata Cristiana)

also complained of manipulation of vote outcomes under the hand-raising
method of voting, but in an interview he described a conscious decision to
tolerate such abuse during the tumultuous 1980s:
[Electronic voting] has already been considered [in El Salvador]. This was many
years ago, around 1985. A committee studied the possibility, and at that time
USAID offered to pay for it, but at that time issues were so complicated – we were
in the war, at times it was necessary to contrive votes, to find a way, in order to move
forward, and it was said that [electronic voting] was not appropriate at the time and
we should wait a while.
By Alvarenga’s estimation, sufficient time has now passed that the Salva-
doran Assembly should adopt electronic voting, but to this point it has not.
It is worth noting that Alvarenga’s PDC was a majority party during the
period when he found ‘‘contriving votes’’ acceptable practice, but currently
it is much smaller and outside government.
Legislators interviewed in every chamber that lacked electronic voting
asserted that outcomes were altered when signal votes, which generally
involve standing, hand raising, or banging on the table (el pupitrazo), were
77
used, and they generally expected that electronic voting would remedy the
problem (interviews with Landazuri and Lucero in Ecuador, Sánchez Bez-
raı́n in Bolivia, Navarro in Colombia, and Guido in Costa Rica). Nicara-
guan deputy Marı́a Lourdes Bolaños (FSLN) confirmed the improvement
under electronic voting on the basis of recent experience:
I think the change is transcendental. Members of the Salvadoran opposition have
told me they want to acquire an electronic system because they regard transparency
as very weak in El Salvador, to the point where the Junta Directiva manipulates
votes. They always overcount, they’re never satisfied. In contrast, we are satisfied
with the votes. We believe there is transparency, we believe there’s efficiency.
That’s important. With manual voting, for all the time it would cost us, now we
have agility. It’s not just about transparency, but agility.
4.3.2.3. Putting Adversaries on Record. Procedural concerns with ob-

struction and accurate vote counting aside, the most common motivation
behind demands for recorded voting among opposition legislators is trans-
parency. The practice forces those who control the legislative agenda to go
on the public record with specific votes to which citizens might object and
whose publication might benefit the opposition (Jones and Hwang 2005).
Statements along these lines from opposition legislators were abundant
(e.g., the Bedregal and Sibaja interviews). Former Bolivian deputy Alfonso
Ferrufino (Movimiento Bolivia Libre) describes the reason majority coal-
ition legislators resist electronic voting as ‘‘the intention and the will of the
representative not to be transparent in his job – to say one thing in public
and do another inside the Congress.’’
A staff member (personal communication) for an opposition COPEI
(Comité de Organización Polı́tica Electoral Independiente) deputy in Ven-
ezuela’s National Assembly portrayed the governing party’s resistance to
electronic recorded voting as less subtle, claiming that state-of-the-art
voting equipment installed with support from a foreign aid program was
vandalized by members of the majority party to avoid having to take res-
ponsibility for their votes. In Panama, an opposition-led proposal to force
the chamber leadership to use the existing electronic voting system was
justified on the grounds that majority-party legislators ducked responsibil-
ity for votes in favor of unpopular tax legislation and motions to grant
immunity from prosecution to government officials charged with wrong-
doing (Tapia G. 2004).
Members of majority parties or coalitions were inclined to dismiss the
importance of electronic voting on the grounds that recorded votes can be
78
requested in any instances either where there is doubt about the outcome or
where enough legislators want to insist on a public record (Acosta, Carva-
jal, and Lucero interviews). Yet opposition legislators objected that those
who control the flow of legislative traffic fail to handle such requests even-
handedly (Devia, Garcı́a Valencia, and Holguı́n interviews). According to
Colombian senator Rafael Orduz,
Orduz: Sometimes, if a group of senators opposes a project and is in the minority,

but it’s in our interest that how everyone voted is known, we can demand
a recorded vote, if we are recognized to speak.
Author: How many do you need to make the demand?
Orduz: One – and it has to be approved by the chamber.
Author: By majority?
Orduz: Of course, and this too can be voted by pupitrazo!
4.3.2.4. Party Mavericks. Making votes public makes it easier for legis-
lators to stake out positions independent from those of their parties.
Recorded votes can serve as means for maverick legislators to ‘‘go public’’
over the heads of party leaders and, in so doing, to establish reputations
either among a target audience of supporters or perhaps nationally. The
rare decisions to hold nominales in systems where anonymous legislative
voting is the norm can illustrate this. According to Costa Rican minority
leader Sibaja (interview),
One sign that there’s going to be a nominal isn’t that the opposition is divided –
that’s no problem. The problem is when the governing party is divided. There was
a famous case here in the early 1970s, having to do with student protests over an
agreement that permitted a transnational company to mine [in a wilderness area]. It
was called the Alcoa Agreement. At that time, the PLN controlled the presidency
and had a big parliamentary majority. One government deputy started the fight.
That deputy himself later became president, but not as a member of the PLN – don
Rodrigo Carazo Odio, who founded the Unity Party, which is governing currently.
He led a group of PLN deputies to break the party line. I think that was the last time
they used a nominal on an important issue, precisely because the government’s
fraccioń divided at that moment. That was thirty years ago. It’s not common.
By Sibaja’s account, public voting on the Alcoa Agreement provided

a mechanism for a deputy with national ambitions, Rodrigo Carazo, to
draw a line in the sand between himself and his party’s leadership. This
suggests that electronic voting, and visible votes more generally, should
encourage independence in legislative voting both insofar as they provide
79
party mavericks with a forum for position taking and insofar as they open
legislators to demands of accountability for their votes from actors outside
the chamber (Bolaños interview).6
In the first years of this century, a similar story was unfolding in Nicar-
agua, where a group of FSLN deputies objected to a deal cut between their
leaders and those of the president’s Liberal Party on a package of consti-
tutional and electoral system reforms. The FSLN dissidents took advan-
tage of the recently adopted electronic voting system to publicize their
rebellion, drawing the ire of loyalists, such as Deputy Maria Lourdes Bola-
ños: ‘‘[The voting records] have been used to make public the divisions
within coalitions. Not for transparency, but to the advantage of those four
deputies who are against the pact. That’s what it has come to’’ (Bolaños and
Urbina interviews). Bolaños evidently saw no boost to transparency inher-
ent in making votes visible. One of the dissidents, Deputy Mónica Balto-
dano, however, was becoming acutely aware of the costs of going public
with her breech of voting unity: ‘‘We broke discipline. . . . So [the party
leadership] ruled that whoever did not accept party decisions could not
aspire to electoral posts. Everyone knew I wanted to run for mayor of
Managua, and this way I couldn’t be nominated. It’s almost certain that
they won’t permit me to run for reelection as a deputy either.’’ As antici-
pated, Baltodano was subsequently barred from nomination for reelection
as deputy, citing as the reason her vote in the Assembly against the electoral
reform law. By 2002 she was out of elected office.
4.3.3. Outside Actors

Other sources of demand for recorded voting are outside the state. By and
large, the public clamor for voting records is modest in the countries where
records are not regularly kept. Legislators in Bolivia and Colombia – even
those who strongly favored recorded voting themselves – described a gen-
eral lack of public attention to individual legislators’ voting behavior (Car-
vajal, Cárdenas, and Holguı́n interviews). Nevertheless, there are pockets
of interest. Organized interest groups – unions, business organizations, and
6
Testing this proposition empirically – for example, by comparing party unity levels in
legislatures with electronic voting against levels on recorded votes in legislatures without
electronic voting – is problematic, because votes that get recorded in the latter may be
biased toward disunity. In the Alcoa vote described by Alexis Sibaja, for example, both
major parties split.
80
farmers’ groups – sometimes monitor legislative voting, even in systems

where no records are kept, and lobby legislators and party leaders to sup-
port their demands (Navarro and Sánchez de Lozada interviews).
More general demand for recorded votes comes from the media, from
academics, and sometimes from watchdog nongovernmental orga-
nizations. During Colombian elections since the late 1990s, Congreso
Visible/Candidatos Visibles, based at the Universidad de los Andes in
Bogota, has solicited background information and policy position state-
ments from all legislative candidates and published all responses on its
Web site, supplementing this material between elections with informa-
tion on partisanship and committee assignments, policy proposals, the
status of legislation, and public statements by legislators. The organiza-
tion also collects the few votes recorded at the individual level in the
Colombian Congress and has been aggressive in lobbying for the adop-
tion of recorded public voting as a matter of standard procedure (Ungar
Bleier 2002).
In Argentina, the Asociación por los Derechos Civiles has pursued a
judicial strategy, filing suit in Buenos Aires municipal court demanding
full public disclosure of all municipal council votes while simultaneously
lobbying at the national level for recorded voting by publicizing contro-
versial legislation on which the public records produced by Congress
do not make it possible to determine how individual legislators voted
(Gonzalez Bertomeu 2004; Clarıń 2004a). In Panama, the local branch of
Transparency International supported opposition-led efforts within the
Legislative Assembly to require that the electronic voting system be used
(Tapia G. 2004). In Mexico, as well, persistent pressure from academics at
the Instituto Tecnologico Autonomo de Mexico (ITAM) during the late
1990s and early years of this decade appears to have hastened the systematic
dissemination of voting records via the Congress’s Web site.
4.3.4. Presidents
Presidential commitment to recorded voting is a product of the specific
political conditions at hand and the goals of a particular president. The rare
circumstances that land the issue at the top of a presidential agenda, how-
ever, may be sufficient to establish recorded voting as standard practice.
As with party leaders, interview subjects dismissed the need for execu-
tives to rely on recorded voting to monitor their legislative allies, on the
grounds that informal networks within legislatures themselves were
81
sufficient for these purposes (Guerra and Holguı́n interviews). Neverthe-
less, presidents may demand recorded voting for other reasons – as a gambit
to enfranchise third-party monitors to offset the inherent advantages of
legislative party leaders, or even out of a genuine desire to increase trans-
parency in the policymaking process.
Many presidents express a generic interest in modernization and effi-
ciency and a willingness to push such demands on a reluctant legislature.
An ironic example is the case of Alberto Fujimori, whose administration is
not generally associated with transparency. Yet Fujimori’s campaign to
modernize the state included an initiative to computerize the Peruvian
Congress, which in turn included the installation of electronic voting
machines (Cevasco Piedra 2000). Although Fujimori’s legislative allies
initially refused to use the equipment, the legislative opposition eventu-
ally succeeded, through the aggressive use of obstructionist tactics, in
forcing the adoption of electronic voting as standard operating proce-
dure (Carey 2003). Thus, Fujimori’s modernizing drive appears, un-
intentionally, to have produced the regular practice of recorded voting
in Peru.
Less inadvertently, and also less successfully, Colombian President
Alvaro Uribe’s first act as president in 2002 was to introduce a broad pack-
age of political reforms, the first element of which was a requirement that
all votes taken in the Colombian Congress be recorded and made public (El
Tiempo 2002). The priority Uribe gave to this procedural detail is remark-
able given that he assumed the Colombian presidency in the midst of a civil
war.7 His stated rationale was that lack of confidence in government insti-
tutions was responsible for the crisis of the Colombian state and that trans-
parency was needed to produce greater accountability among elected
officials.8 Facing congressional resistance to his proposal in 2002, Uribe
put his reform package directly to voters in a fifteen-point referendum the
next year. The public voting provision received 94 percent support
among votes cast, but only 24.7 percent of eligible voters participated. In
Colombia, referenda require 25 percent participation to be valid, so Uribe’s
proposal failed, and most votes in Colombia remain invisible to those out-
side the Congress.
7
Uribe’s inauguration ceremony itself was subject to a mortar attack.
8
It is worth noting that Colombia’s previous president, Andres Pastrana, in the name of
transparency, had also tried to pass a package of reforms that included the requirement of
public voting. The proposal died in the face of legislative opposition.
82
Effects of Recorded Voting
4.4. Effects of Recorded Voting

Are the motivations and concerns of these actors who care about the move to
recorded voting justified? What impact does recorded voting have on legis-
lative representation? The most obvious effect is an increase in transparency
and greater opportunities for actors outside the legislature to exert pressure
on elected representatives. The discussion of monitoring also suggests that
publicizing votes should weaken party leaders’ influence over legislators.
This section presents evidence confirming both of these propositions.
4.4.1. Transparency
When asked the open-ended question, ‘‘What effects, if any, does recorded
voting have on legislative representation?’’ most legislators mentioned an
increase in transparency.9 The term, however, is sufficiently general (and
such talk is sufficiently cheap) that it is worth spelling out more explicitly
what it entails. In its crudest sense, the transparency afforded by recorded
voting is a check against the ability of legislators to lie outright about the
policies they have supported or opposed inside the chamber. None of my
interview subjects confessed to having perjured themselves in this manner,
but claims that their colleagues had were common.10 Colombian senator
Rafael Orduz (interview) was willing to name names:
I’ll give you an example, having to do with a particular part of a recent tax reform. In
public – I’m talking about on television – the president of the Liberal Party, Luis
Guillermo Veles, said he was against it. In the vote, inside the Senate, he voted in
favor. There was no TV and no recorded vote, but nobody has called him on it in
public. So publicly, he continues to position himself as if he had opposed the article
I’m talking about.
Orduz’s point was that a recorded vote on the tax measure in question would
have offered a deterrent against the obfuscation of which he accuses Senator
Veles, or else have provided incontrovertible evidence with which any of
Veles’s opponents (presumably including Orduz) could expose his perfidy.
Another Colombian senator offered a similar assessment with respect to
why the Colombian Senate never used an electronic voting system that
been installed two years before: ‘‘They say it has technical problems, but
9
In legislatures that do not regularly record votes, I asked, ‘‘What effects, if any, do you
expect recorded voting would have on legislative representation?’’
10
I am still not sure how I managed to overselect honest politicians to such an extreme.
83
this is just an excuse because they don’t want votes to be public. Too many
senators are afraid they will lose votes if they are all made public’’ (Blum
interview). I pursued the same issue later, at a meeting attended by a group
of senators and representatives, asking why neither chamber of the Colom-
bian Congress used its electronic voting system:
Anonymous senator: ‘‘There were problems. They didn’t work.’’
Author: ‘‘Why not? Technical problems?’’
Anonymous senator: ‘‘Well, politico-technical problems.’’
All legislators: [Big laughs all around the table.]
Beyond their jaded perspective, these Colombian examples reflect the

belief that the transparency provided by visible votes acts as a deterrent to
bad behavior – to lying about votes or voting against constituent interests –
by legislators. Most of those interviewed for this project also mentioned
more general positive effects of transparency, suggesting that voting
records are a basic mechanism for transmitting information about legisla-
tive decision making and that this information is a public good. Without
rehearsing each such statement, Ecuadorian Alexandra Vela (Democracia y
Progreso) provides the general flavor:
The mechanism for rendering accounts doesn’t function if there’s no way to verify
the votes. Why? Because from this election we have the obligation to present
a legislative program. So, we notarize, we go to the notary and we publicly say this
is the program. But citizens request accounts from their delegates as to whether this
is the program that was presented and you don’t know how each one voted. So, the
process of rendering accounts demands that there be voting of this kind. Also, for
knowledge and for learning, as a pedagogical matter for citizens who don’t un-
derstand and can’t see how the mechanisms of a democracy function. It’s important
for them to see it.
4.4.2. Direct Effects on Policy Decisions

Some of the general claims about transparency point to education and
information (Ferrufino interview). Others suggest that the threat of sanc-
tion from voters deters legislators from shirking constituents’ interests in
their voting behavior, which in turn suggests the stronger claim that
recorded voting can change actual policy outcomes (Blanco Oropeza in-
terview). Testing such a claim systematically is inherently difficult, of
course, because in any given case the outcome realized must be compared
against the counterfactual – the outcome that would have been realized in the
absence or presence of recorded voting.
84
One example, drawn from the Texas state legislature, suggests such an
effect for recorded voting. In April 2003 the state House of Representatives
was considering a motion to kill a widely popular proposal to require
legislative candidates to disclose the sources of their campaign contribu-
tions. The motion was initially put to a nonrecorded vote and appeared to
be headed toward passage (which would have killed the bill); however,
‘‘When a recorded vote was requested, the scoreboard changed completely
and the motion to kill the bill failed’’ (Dallas Morning News 2003b).
Multiple independent interviews conducted in Peru pointed toward
another case where recorded voting changed an important policy outcome
(Ması́as, Pease, De Althaus, and Ortiz de Zevallos interviews). At issue was
a proposed change in the electoral system for the 2001 election. Incumbent
legislators had been elected from a single national district. The proposed
reform would divide Peru into multiple electoral districts (circunscripciones)
defined by departmental boundaries. Despite popular support for the idea,
many legislators were skeptical about altering the rules under which they
had, by definition, been successful. Congressman Henry Pease (Union por
el Peru) provided the most compelling account of recorded voting’s effect
on the outcome:
[This reform] obviously was not good for small parties, or for those that knew
that, after the way they had governed, they were going to be small parties. There
was a lot of tension when this issue was put to debate, with strong public opinion
in favor, and a bunch of legislators demanded, based on an article of the rules,
that the vote should be secret. I wasn’t in Congress at the moment because I was
sick in the medical clinic, so I saw on television the impact, above all, of public
opinion. I was in the clinic at least from 6:00 or 7:00 P.M. on, in a room watching
the TV, and the nurses were coming in – not to look after me, but to watch the
TV and express their indignation at what was happening, because as soon as they
saw that it was going to be a secret vote they said ‘‘It’s going to lose’’ and in fact,
they didn’t get enough votes to get rid of the national district. This led to
a mobilization and to demands of all sorts and criticisms of all sorts and allowed
us to force, a month later, another proposal, and vote on it . . . and finally,
even though it was a much more radical bill, it was accepted because of public
pressure.
4.4.3. Do Citizens Pay Attention?

Pease’s comments suggest a critical issue with respect to the prospects for
recorded votes to act as mechanisms for legislative accountability in the
manner that they do in the United States – whether voting records find
their way into the mainstream of political discourse. To put the matter
85
more bluntly, even if votes are visible, does anyone look? The Peruvian
fieldwork for this project was conducted in May 2001 – a period of intense
politicization in that country, after the fall of the Fujimori regime, and in
between the two rounds of the election that produced the presidency of
Alejandro Toledo. Congress, which had produced the interim president
Valentı́n Paniagua, and which was conducting investigations into the spec-
tacular corruption charges against members of the Fujimori administra-
tion, was in the national spotlight. In this context, the Peruvian media
reported intensively on congressional voting records. Newspaper articles
reproduced roll calls (Diez Canseco 2001; La Gaceta 2001), and television
talk shows focused on motivations (De Althaus interview). Recorded voting
in Peru appeared to have, in very short order, established itself at the core of
political discourse.
One should be cautious, however, about generalizing too quickly from
the experience of Peru in 2001, which was extraordinary on a number of
counts. The early experience of Nicaragua with recorded voting stands in
contrast to Peru. Field research in Nicaragua was conducted in August
2000, and electronic voting had been adopted as standard operating pro-
cedure only eight months earlier, in January. Deputy Carlos Hurtado
(Accion Conservador) described the status of the voting records this way:
Hurtado: Despite the fact that votes are recorded, they aren’t widely known
among the people, except when a particular issue becomes decisive at
election time.
Author: Is this a process that is beginning?
Hurtado: Yes, it’s happening. It requires that the electorate, the political analysts,
have the record. It’s not so simple to create this record because it
requires a certain infrastructure, a certain culture, a certain systematiza-
tion. There’s no independent center that keeps a record of the votes. In
the United States, there is – there are lots that keep complete records of
the details. That’s more sophisticated. I think eventually we’ll get there,
but certainly as of now it’s not so easy. At least it’s known when a certain
deputy takes a certain position.
Individual votes are, indeed, known within the Nicaraguan Assembly

because its electronic voting system includes a large screen indicating each
deputy’s position on each motion, yet they are not widely known beyond
the Assembly hall because the records are not published, nor is it easy for
would-be observers from outside the Assembly to obtain them. Deputy
Jorge Samper (Movimiento Renovacion Sandinismo) pointed toward the
status of legislative staff – and specifically of civil service protection of
86
government employees – in explaining the obstructed flow of information

about what goes on inside the legislature to sources on the outside. In
Nicaragua, according to Samper, legislative employees, lacking civil service
protection, are reluctant to release voting records out of fear that any
discomfort those published records generate among powerful politicians
could cost the staff members their jobs:
Right now some [members of the legislative staff] still resist publishing things
without someone giving them orders to do it. Sometimes they’re a bit afraid. . . .
It’s important that there be career officials who will be the institutional memory
who attest to what’s done and that it be published – that there be a guarantee,
a nonpartisan guarantee, independent of who’s in the government or who has the
majority in the parliament, in order to provide real, effective, and concrete infor-
mation, without of any sort of fears.
4.4.4. Visible Votes and the Quality of Representation

Beyond the accounts of legislators themselves, it is difficult to estimate the
effect on transparency of recording and publishing legislative votes. The
Corruption Perceptions Index (CPI), calculated by Transparency Interna-
tional for most countries in the world, measures perceptions of corruption
generally. The CPI is not legislature-specific, but it is a widely recognized
measure of political transparency and allows cross-national comparison.
Figure 4.1 plots the square root of the average number of votes recorded
and made public per year, by country, against the 2004 CPI. I use the
square root of the number of votes per year on the logic that there are
diminishing returns to the information conveyed through visible votes
(positive correlation .55, or .37 when dropping the United States and
retaining just the Latin American cases) between the log of public votes
per year and the CPI.
Whether because legislative transparency reduces corruption, or be-
cause some combination of factors that produces legislative transparency
also contributes to cleaner government in general (or, more likely, some
combination of these effects), perceptions of corruption tend to be lower in
countries where legislative votes are visible.11
11
I found no relationship between public votes and ‘‘confidence in Congress,’’ as measured
by the Latinobarómetro during the late 1990s through early 2000s. To the extent that the
crafting of legislation resembles sausage making, per Bismarck’s famous observation, vis-
ibility may enhance accountability and deter corruption without necessarily increasing
public affection for Congress.
87
Figure 4.1 Recorded votes per year (square root), by country, and 2004 Trans-
parency International Corruption Perceptions Index.
The experiences of specific countries demonstrate that individual legis-

lators’ voting records can attain political salience in Latin America and can
do so relatively quickly, as in Peru. The broader pattern is confirmed as well
by Brazil and Chile, with longer experiences with recording votes. On the
other hand, it is not automatic that ‘‘If you record it, they will come.’’ MMD
elections in Latin America mitigate incentives to monitor individual voting
behavior, and resistance to recorded voting from powerful legislative actors
also discourages forces otherwise inclined to publicize voting records.
4.4.5. Partisan versus Individualized Legislative Representation

Much of the discussion so far has suggested that recorded voting can under-
mine discipline in legislative parties and coalitions. Discipline implies the
ability on the part of party leaders to compel legislators to vote contrary to
their immediate preferences or contrary to preferences induced through
their electoral connection with constituents. Recording votes and making
them public increase the costs to legislators of voting that is disciplined in
this sense. The fact that party leaders, in interviews, supported recorded
voting less than other legislators is consistent with this interpretation.
88
This analysis cannot test the hypothesis that recorded voting weakens
discipline, for the obvious reason that no data exist and no levels of party
voting unity can be measured when voting records do not exist, but the
interviews support this proposition. Deputy Mónica Baltodano, the FSLN
dissident sanctioned for voting against the party, described strong public
support for independence from absolute party discipline: ‘‘There is a great
tolerance [of indiscipline] among citizens and Nicaraguan society, which
can be demonstrated in polls and studies. But within the political institu-
tions, there’s great intolerance, above all within the party.’’ A vignette
provided by Peruvian congressman Carlos Blanco Oropeza (Cambio 90 –
Nueva Mayoria) illustrates the expected effect among party leaders of
recorded voting on discipline in more colorful terms:
I’ll never forget when, in 1998, I was vice president of the Congress, and we hosted
some German legislators, and naturally one of the things we did was to visit the
facilities of the Congress. I accompanied them to the chamber, the place where we
meet, and right there is the screen for the public votes, and I’ll never forget the
German legislator – who was a leader of his party – and he said to me, ‘‘Is the vote
public?’’ So I explained, and he said, ‘‘You guys are crazy. How can you control the
members of your party? Everyone has to vote how the party votes.’’
Blanco’s German colleague was referring to the practice of signal voting

in the Bundesrat, a practice with which Blanco himself was familiar because
the Peruvian Congress had relied almost exclusively on signal voting until
modernization, including the installation of electronic voting equipment,
in the late 1990s ushered in recorded voting. Blanco confirmed that
recorded voting increased the willingness of individual legislators to resist
leadership directives (Blanco Oropeza interview).
Whether the potential for recorded voting to undermine party discipline
is realized depends on whether citizens come to regard voting records as
salient in evaluating legislator performance, whether they are willing to
reward independence from parties in the pursuit of some other conception
of constituent interests, and whether these forces supersede whatever other
tools party leaders retain to enforce discipline. Interview subjects con-
tended that voting records encourage responsiveness to citizens in legisla-
tors’ immediate electoral constituencies, even at the cost of loyalty to the
national party, because voters are prepared to weigh either regional inter-
ests against those of the national party (A. Alvarenga interview) or sectoral
interests (Ması́as interview).
The priority of individual over partisan accountability to the electorate
was a theme repeated in interviews in one country after another, even where
89
recorded voting has made no progress toward adoption. In Venezuela, for

example, the same antiparty chorus was echoed by both opponents and
supporters of President Hugo Chavez. Former Constituent Assembly dep-
uty Ricardo Combellas described the philosophy behind the new constitu-
tion adopted in 1999: ‘‘We wanted to eliminate partyarchy – to eliminate it
constitutionally, but in terms of norms, for the representative to respond
more directly to the wants and needs of his constituents. Responsibility in
parliament is personal – the Constitution says so – not to respond to a party
but to one’s constituents.’’ William Tarek Saab (Movimiento Quinta
Republica), a leading member of President Hugo Chavez’s party in both
the Constituent Assembly and the current legislature, concurred: ‘‘I think
that here you have to listen to the voice of the people. One of the require-
ments for this is the vote of conscience. You pay attention only to your
conscience.’’ To this point, of course, whether votes in the Venezuelan
National Assembly are regularly being cast according to the demands of
conscience – perhaps even independently from party lines – must remain
a matter of faith, because of the failure of the Assembly’s leadership to use
the electronic voting system that is in place.
4.5. The Trend toward Visible Votes and Its Limits

Recorded voting is a controversial reform, resisted by powerful legislative
actors. Its principal advocates are opposition legislators and dissidents
within governing coalitions, as well as nongovernmental organizations
and journalists for whom government transparency is a priority, and the
odd academic. Its main opponents are the leaders of governing parties. In
short, those inclined to oppose recorded voting determine who sets rules of
legislative procedure. On the other hand, the gradual spread of electronic
voting in many Latin American legislatures over the past decade or so
indicates that the barriers are not insurmountable and that demand for
visible votes sometimes prevails. Procedural rights granted to legislative
minorities may provide some leverage for those demanding the regular use
of recorded voting, as was the case in Peru. In Panama, and in the state of
Texas, pressure from the media preceded the adoption of visible voting in
legislatures. The Colombian case suggests that supporters of recorded
voting might occasionally find allies with clout in the executive branch,
and in Argentina, transparency advocates are looking to the courts.
Perhaps most important to the overall trend is that recorded voting as
standard practice appears to be subject to a sort of ratchet effect. Once the
90
The Trend toward Visible Votes and Its Limits
practice is adopted, it is difficult to backslide. According to Peruvian con-

gressman Blanco Oropeza, ‘‘Once it’s done, you’re not going to change it.
It’s not going to change because even if the general public doesn’t pay much
attention, the journalists do. The journalists and the other politicians, too,
because they are the ones who get accustomed to using this information.’’
In another interview, political commentator Jaime De Althaus, also from
Peru, summed up the matter even more categorically: ‘‘[Recorded public
voting] has had its own inertia. . . . It’s an almost inevitable consequence
that runs according to its own logic.’’
The point is that, although opponents of visible voting may be able to
keep the issue off the reform agenda quietly for extended periods of time,
once the practice is established, a move to eliminate it would be difficult to
defend publicly. In this sense, the empirical evidence from this chapter
suggests that the case for visible voting, as well as the conception of legis-
lative accountability that goes with it, resonates widely. In spite of scholarly
appeals for responsible party government, or even the arguments of Rous-
seau and Burke about legislative detachment, citizens want to know how
their representatives vote – or at least representatives are reluctant to sug-
gest otherwise.
This is not to say we should necessarily expect legislative voting records
to assume in other political contexts the central role they play in U.S.
legislative politics. In particular, MMD elections dilute the incentives for
individual politicians to use incumbents’ voting records as ammunition in
electoral battles. Nevertheless, many actors are motivated to promote vot-
ing records, technological advances are on their side, and backsliding on
this matter is improbable. Recorded voting, therefore, should become
more common, and voting records should grow increasingly salient to
political debate and increasingly central to the accountability relationships
between legislators and voters, even if they never attain the prominence
they have in the United States.
91
5
Counting Votes
5.1. Party Voting Unity and Collective Accountability

When votes are visible, what can we learn from them about legislative
accountability? Citizens can learn whether specific representatives have
pursued their interests in motions put to a vote on the floor – information
that can provide a basis for individual legislative accountability. Even apart
from the policy substance of the votes themselves, voting patterns can tell
us about prospects for collective accountability, because collective account-
ability requires that groups of legislators vote in a unified manner to shape
outcomes on the floor.
I propose various measures of voting unity and success among groups
of legislators. Because parties are the ubiquitous organizers of legislative
work, the groups on which I focus primarily are parties, so unless otherwise
noted, voting unity and related terms refer to unity among members of the
same party. The measures of unity, however, can be equally well applied to
any other group, such as coalitions that encompass more than one party or
legislators from a particular region, sex, race, or religion – any character-
istic of interest for analysis.
Why should we care about party unity in legislative voting? First, leg-
islative votes are the means by which major public policy decisions are
ratified in all democracies. Voting behavior is of intrinsic interest because
the stakes are high. Second, political parties are potentially important in-
formation conduits to citizens. Parties can pledge to support comprehen-
sive policy platforms on which individual politicians cannot credibly claim
to have much impact. Whether voters can know what they are getting in
elections depends partly on legislative voting unity. If the voting behavior
of a party’s legislators is unrelated to the positions in its national platform,
92
Party Voting Unity and Collective Accountability
then the party’s label has no informational value. Third, unity affects the
ability of parties to win votes and shape policy. Unity determines whether
governments can act decisively or, by contrast, whether each legislative
decision requires separate deliberation and the construction of a distinct
support coalition. In this sense, party unity is linked to the ability of parties
and governments to deliver the promises in their platforms (Bowler,
Farrell, and Katz 1999).
I focus on three main characteristics of how parties vote: how consis-
tently their members take the same position on the motions on which they
vote, how much they win, and how frequently their losses might have been
avoided but for breaches of voting unity. Voting together matters not just
because mobilizing its full voting capacity can help a party win floor votes
and so promote its supporters’ interests. Voting together (or failing to do
so) also sends information to citizens about the party’s policy positions and
its level of commitment to them. When votes are visible, cross-voting
within a party or failure to mobilize its members on a given measure blurs
the party’s brand name. A party that mobilizes its potential and votes in
a unified manner, by contrast, fortifies and clarifies its reputation. Thus, the
degree to which copartisans vote together is relevant to accountability both
through the delivery of wins and losses and through its communicative
content.
The indices developed in this chapter measure raw levels of party unity
and are agnostic as to whether the source of unity is like-mindedness among
legislators (cohesiveness) or pressure from party leaders (discipline). Raw
levels of party unity matter in their own right. Low unity reduces the com-
municative value of party labels to voters, and high party unity increases it,
regardless of whether the source is cohesiveness or discipline. Low unity
reduces the ability of parties to deliver their policy promises through legis-
lation, and high party unity increases it, regardless of whether the source is
cohesiveness or discipline. Indices of party unity that can be deployed across
systems and across parties describe variance in a common metric that makes
cross-national comparison possible. As this chapter demonstrates, party
unity varies a lot, both across political systems and among parties within
systems. The next chapter uses the indices developed here to shed some
light on the relative importance of cohesiveness versus discipline in account-
ing for the levels of party unity observed empirically and to test propositions
about the sources of high and low party unity.
Party unity is a necessary condition for collective accountability, not its
guarantee. As Chapter 2 highlights, when voters are unable to demand
93
Counting Votes
accountability from national party leaders, highly unified parties can be

turned into cartels to monopolize state resources on behalf of narrow sets
of beneficiaries or to pursue policies contrary to those advertised in elec-
toral campaigns, to voters’ widespread chagrin (Coppedge 1994; Stokes
2001). Yet without unity, parties can neither communicate a coherent col-
lective reputation to voters nor credibly claim to deliver on policy plat-
forms that reflect that reputation. As long as party unity is a condition for
collective accountability, even if not its equivalent, it is important to de-
velop a common metric to compare unity across political systems.
The first part of this chapter develops some indices I use to measure
voting unity, discusses their properties, and illustrates them by applying
them to votes in hypothetical legislatures as well as to some empirical
examples. Then I present recorded vote data collected for this project from
legislatures in nineteen countries and apply the measures to describe par-
tisan voting unity across these chambers statistically and graphically.
5.2. Measures of Voting Unity and Success
5.2.1. RICE and UNITY Scores

I rely on a variety of indices built from the voting record. The indices are
summaries of information from across multiple votes. Their most basic
building blocks are the aye and nay votes, abstentions, and nonvotes cast
by individual legislators. These are used to construct party voting unity
scores for individual votes, which in turn are aggregated into indices that
describe patterns across sets of votes.
The first measure is familiar to legislative scholars and is based on the
measure of unity developed by Stuart Rice (1925) more than eighty years
ago. RICE scores reflect levels of cross-voting among members of the same
party on a given vote and are calculated as:
RICEij ¼ jAYEij NAY ij j for party i on vote j;
where aye and nay are calculated as proportions of those voting either aye or
nay, and so sum to 1.0. The RICE score can range from zero (equal numbers
vote aye and nay) to one (all members who cast votes vote together).
One limitation of RICE is that it does not account for nonvoting, levels
of which are substantial in most legislatures. For example, if a party has 100
members, 60 of whom cast affirmative votes on a measure and 40 of whom
94
Measures of Voting Unity and Success
abstain or otherwise do not vote, RICE regards this event as perfect unity
(RICE ¼ 1.0), equivalent to if all 100 members of the party voted aye (or
nay). Intuitively, these are two fundamentally different events, the differ-
ence between which could obviously affect the vote outcome. This suggests
some type of measure that is sensitive to whether copartisans vote, as well as
to how they vote. For this, I propose UNITY, which captures the extent to
which a party exercises its decisive capacity on a given vote:
UNITY ij ¼ jAYEij NAY ij j for party i on vote j;
where the proportions are calculated as shares of all members of party i in

the legislature.
Like RICE, UNITY can range from zero to one, but it ‘‘dips’’ more
easily than RICE, taking the minimum value either if those legislators who
do vote split evenly between aye and nay (like RICE) or if no members cast
decisive votes. It takes the maximum value if all members cast decisive (i.e.,
aye or nay) votes in the same direction, and it falls between these extremes
when either decisive voters within the party divide against each other, or
some members vote while others do not, or both. In effect, UNITY is
a cousin to RICE, but it is discounted according to the rate of nonvoting
in the group. For an illustration of RICE and UNITY scores from a hypo-
thetical assembly, see Appendix 1 at the end of this chapter.
5.2.2. RICE and UNITY Indices

Because I am aiming to discern general characteristics of party groups, it is
convenient to aggregate the vote-specific measures of voting unity into
indices that summarize, for each party, the overall tendency toward unity
across all the recorded votes in a given legislature. I aggregate at the level of
legislatures because this is the largest period for which some important
characteristics of party groups (e.g., share of seats, government or opposi-
tion status) are constant.
One problem with simply averaging voting scores to create indices is
that many votes in most legislatures are lopsided, either because they are
taken on matters of consensus across parties or on matters unimportant
enough to attract any opposition, or because their outcome is obvious
ahead of time and the losers may choose not to register their opposition
formally through their votes. To correct this problem, when aggregating
parties’ RICE and UNITY scores across votes into indices, I weight the
95
Counting Votes
score from each vote according to how closely the vote was contested. The
procedure for calculating weighted RICE (WRICE) and UNITY (WUNI-
TY) indices is described in Appendix 2.
5.2.3. Winning, Losing, and Voting Unity

Mobilizing legislators and voting together matter to parties’ collective
reputations and may also matter to whether parties win or lose votes, but
winning and losing can also be observed directly, along with the relation-
ship between mobilizing votes, cross-voting, and voting success rates. I
calculate for each party, i, on each vote, j, whether it wins, WINij, which
is simply an indicator of whether the plurality of party i’s members voted on
the winning side. The details of how this legislative success index is created
are described in Appendix 3.
Finally, whereas RICEi and UNITYi reflect the extent to which parties
project, through legislative voting, collective reputations, my last pair of mea-
sures is based on the intuition that unified parties more effectively influence
policy than disunified ones. RLOSERij and ULOSERij reflect whether a party
suffers a loss on a given legislative vote because, respectively, of cross-voting
or of a failure to mobilize its full voting capacity. RLOSERij takes a value of 1
if party i loses on vote j even though, given how all other legislators voted, party i
could have won had all its voting members voted together. ULOSERij takes
a value of 1 if party i loses on vote j even though, given how all other legislators
voted, party i could have won had it mobilized its full voting capacity behind
its preferred outcome. Details on the calculation of these indices, as well
as illustrations from a hypothetical legislature, are in Appendix 4.
5.3. The Silence of Nonvotes
5.3.1. What Do Nonvotes Mean?

Because they take account of nonvoting, and so capture the degree to which
a party mobilizes its full potential vote behind its preferred positions,
UNITYi and its cousin, ULOSERi, might appear to be more comprehen-
sive and reliable reflections of the extent to which a party translates its
legislative representation into influence over legislative decisions than
RICEi and RLOSERi. Yet there is a serious potential shortcoming with
these indices because how one interprets nonvoting is more complex, both
mechanically and strategically, than how one interprets cross-voting.
96
The Silence of Nonvotes
Some studies of recorded votes seek to interpret the motivation behind

nonvotes, in order to infer whether they likely represent breaks with party
discipline – for example, if legislators were present for some votes in a ses-
sion but not others (Haspel, Remington, and Smith 1998; Ames 2002).
This approach implicitly attributes analogous meaning to nonvoting and
to voting, regarding each as an equivalent action for the purposes of mea-
suring party voting unity. Under most conditions, however such an ap-
proach can mismeasure the effects of nonvoting (Jones and Hwang 2005).
On most votes in most assemblies, threshold for approval is set in relative
terms, as a proportion of votes cast (subject to a quorum). But for some votes –
and in a handful of assemblies, for all votes – the threshold for approval is set
in absolute terms, as a proportion of the full membership. The effect of
nonvoting on outcomes depends directly on such rules, and absolute thresh-
olds create particular challenges for the voting unity indices developed in this
chapter. These issues are discussed in Appendix 5. Going forward, my default
approach is to treat nonvotes according to their effects on vote outcomes.
Because of the ambiguity absolute-threshold rules generate, however, I rep-
licate all the quantitative analyses in the book dropping the absolute-majority
cases: Nicaragua, Guatemala, and Russia. Where doing so affects results, I
report both specifications and discuss the differences in interpretation.
5.3.2. Nonvoting Equilibria

Beyond the different mechanical implications of nonvotes under absolute-
versus relative-majority threshold voting rules, nonvoting also carries po-
tentially important strategic ambiguities under either type of rule. Parties
may tolerate nonvoting by members who could have been mobilized, if
necessary. Leaders may strike agreements with rank-and-file members
within their own parties to tolerate nonvoting as long as preliminary head
counts suggest nonvotes will not be pivotal (King and Zeckhauser 2003).
Alternatively, they might strike agreements with leaders of other parties to
‘‘match’’ nonvotes that offset each other across party lines, so as not to
affect the overall outcome. A nonvoting equilibrium arrangement might be
advantageous both to party leaders, as a means of hiding displays of internal
dissent, and to individual legislators, both when they prefer not to support
their party’s line and when they are merely beholden to other commitments
besides attendance and voting on the floor – for example, to committee
work, to constituency service, or even to professional or personal obliga-
tions outside the legislature. The existence of such agreements is asserted in
97
Counting Votes
various Latin American legislatures in interviews conducted during the

course of research for this project. To the extent that such agreements
represent equilibria within or across parties not to mobilize their full voting
potential, and that leaders could mobilize their legislators if necessary, then
their party is potentially more unified than the observed voting record sug-
gests, and the validity of UNITYij and ULOSERij as measures of party
voting unity are dubious.
Consider the illustrations of party voting unity in Argentina and New
Zealand in Figures 5.1 and 5.2. In the figures, WRICEi indices are repre-
sented by the height of each bar on the y axis. Each party’s share of seats is
represented by its width on the x axis. In each case, WRICEi is quite high,
whereas WUNITYi is substantially lower, with the decline of similar
magnitude across parties, indicating little cross-voting, but suggesting
the possibility of matched nonvoting equilibria. The implications for the
win-loss-based indices are striking. For example, Argentina’s governing
Peronist Party (PJ, or Partido Justicialista) was on the winning side in 95
percent of recorded floor votes, but in all of its losses it suffered cross-
voting that, if reversed, would have flipped the outcome in the Peronists’
favor (RLOSERi ¼ 5 percent). The main opposition Radical Party (UCR),
by contrast, lost 85 percent of recorded votes. None of these defeats is
attributable to cross-voting (RLOSERi ¼ 0), but the ULOSERi measure
suggests that 30 percent could have been reversed had the UCR mobilized
its full cohort behind the party’s position. Could the opposition have pre-
vailed in this manner? More likely, had the UCR increased mobilization,
the Peronists could and would have countermobilized, summoning more
votes to the floor (raising WUNITYi for both, and driving the UCR’s
ULOSERi index down). The UCR’s inflated ULOSERi, that is, might well
be a mere reflection of a nonvoting equilibrium across parties – an equi-
librium also reflected in the even drop-off from WRICEi to WUNITYi
indices. The data from New Zealand’s parliament of 1993–94 suggests
a very similar potential relationship between the National and Labour
parties, with an even larger spike in the latter’s ULOSERi index.
The patterns from Argentina and New Zealand suggest equilibria across
governing and opposition parties whereby the former tolerate nonvoting
provided that less-than-full mobilization does not cost them victories, and
the latter tolerate nonvoting, aware that full mobilization would only trig-
ger countermobilization by governing parties. But not all legislatures exhib-
it similar patterns. Consider the pattern in the early months of Alejandro
Toledo’s administration in Peru, from August to October 2001, shown in
98
Figure 5.1 WRICE, WUNITY, RLOSER, and ULOSER indices for Argentina,
1995–97.
Figure 5.3. The drop-off from WRICEi to WUNITYi was more pro-
nounced for President Toledo’s Peru Posible (PP) party than for others,
and it cost the government party victories on about 3 percent of all
recorded votes – a higher rate than for any other party.
Along the same lines, consider the voting record of the Czech Republic’s
parliament during its 1996–98 term. In Figure 5.4, the parties of the gov-
erning coalition are white and all others shaded various hues of gray. There
is the familiar drop-off from WRICEi to WUNITYi indices, and the rise
from RLOSERi to ULOSERi, as the latter capture nonvoting as well as
cross-voting. But the rise in ULOSERi is spread across all parties and is
more pronounced within the governing coalition than outside it. In short,
patterns of these indices across countries suggest that nonvoting equilibria
are possible but not uniform. Moreover, the combined patterns of
WRICEi, WUNITYi, RLOSERi, and ULOSERi across parties can detect
signs of such equilibria in some cases and help rule them out in others.
99
Counting Votes
Figure 5.2 WRICE, WUNITY, RLOSER, and ULOSER indices for

New Zealand, 1993–94.
Furthermore, even where nonvoting equilibria exist, it is not obvious

that these ought to be regarded as cases of party unity on par with those in
which votes are fully mobilized. When rank-and-file legislators have other,
more pressing, priorities than getting to the floor to cast votes in line with
their copartisans, the observed level of mobilization represents an intrinsic
level of support for the party’s position. Leaders might be able to increase
mobilization beyond that point in a pinch, but to do so they might also have
to twist arms and otherwise expend political resources. By this interpreta-
tion, the very existence of nonvoting equilibria can reflect a lack of common
purpose within parties.
Voting records tell us, among other things, who does not vote, but
nonvoting is a more ambiguous act than casting an aye or a nay. In part,
this is because the mechanics of voting can vary across legislatures. Where
the threshold to approve a measure is set in absolute terms, the effect of
nonvoting is substantively different from where the threshold is relative.
100
Figure 5.3 WRICE, WUNITY, RLOSER, and ULOSER indices for Peru, 2001.
Under absolute-threshold rules, I treat nonvotes as nay votes, but I also

replicate all analyses dropping the absolute-majority cases, and the results
reported in this book do not change. In part, nonvotes are ambiguous
because nonvoting may reflect strategic behavior that reflects layers
of agreements among party leaders and between leaders and their
rank-and-file members. The mobilization-based indices, WUNITYi and
ULOSERi, are particularly sensitive to nonvoting and, therefore, to
nonvoting-equilibria.
5.3.3. Small Parties

One last set of issues associated with the voting unity measures employed in
this book focuses on applying the measures to very small parties. WRICE
and ULOSER are not calculated for single-member parties, and RLOSER
is not calculated for parties with fewer than three members. These matters,
101
Counting Votes
Country: Czech Republic. Legislature: 1-Jul-96 Country: Czech Republic. Legislature: 1-Jul-96
Weighted UNITY Index

Weighted RICE Index
1
1
.75
.75
.5
.5
.25
.25
0
0
.2 .4 .6 .8 1 .2 .4 .6 .8 1
Proportion of Seats in Chamber Proportion of Seats in Chamber
ODS ODA KDU-CSL ODS ODA KDU-CSL

SPR-RSC Ind CSSD SPR-RSC Ind CSSD
LB LB
Country: Czech Republic. Legislature: 1-Jul-96 Country: Czech Republic. Legislature: 1-Jul-96
0 .05 .1 .15 .2 .25
0 .05 .1 .15 .2 .25

ULOSER Index
RLOSER Index
.2 .4 .6 .8 1 .2 .4 .6 .8 1
Proportion of Seats in Chamber Proportion of Seats in Chamber
ODS ODA KDU-CSL ODS ODA KDU-CSL

SPR-RSC Ind CSSD SPR-RSC Ind CSSD
LB LB
Figure 5.4 WRICE, WUNITY, RLOSER, and ULOSER indices for the Czech
Republic, 1996–98.
as well as corrections for bias inherent in RICE and UNITY scores for
small parties, are discussed in Appendix 6.
5.4. Data on Recorded Votes

As discussed in Chapter 3, the availability of recorded votes in many legis-
latures, particularly in quantities that facilitate quantitative analysis, is lim-
ited. My main criterion for including recorded votes from a given legislature
in this study was simply whether I could get data. In this sense, data collec-
tion resembled a Drunkard’s Search, with the most energy devoted toward
legislatures where there was sufficient transparency (enough light) that I
might collect some recorded votes. This includes all the cases from the
analysis of transparency (see Table 3.1) that yielded usable recorded vote
data, plus all others from which I was able to collect data. Some further
background on data availability and collection is in order here.
102
Data on Recorded Votes
The study of recorded legislative votes is well established in the United

States, and every recorded vote taken in the U.S. Congress throughout its
history is publicly available in machine-readable format.1 Studies of
recorded votes outside the United States are rare, in part because most
legislatures record very few votes, or none, at the level of the individual
legislator. Where votes are recorded, collecting data often requires travel-
ing to the country, navigating bureaucratic and political hurdles, and tran-
scribing hard copy documents into machine-readable format. This is an
extremely labor-intensive proposition (see the Chapter 3 Appendix). Schol-
ars focusing on a variety of countries have undertaken this work, generally
in piecemeal fashion, producing legislative voting studies for a single coun-
try or just a few countries at most.
Research for this book entailed determining, for Western Hemisphere
democracies, whether recorded votes are available and collecting samples
of votes large enough to allow for quantitative analysis wherever possible.
In most cases, this required field research to collect the data, or at least to
determine that no recorded votes exist. In some cases, I was able to hire
in-country research assistants. Outside the Americas, data collection
confronted bigger obstacles. Language barriers; limited contacts among
politicians, academics, and legislative staff; and finite time and travel bud-
gets prohibited original data collection for the most part.2 To acquire data
under these constraints, I identified academics from other countries who
had recorded vote data and who might be willing to share or trade. This
approach yielded datasets from Australia, Canada, the Czech Republic, the
French Fourth Republic, New Zealand, Poland, and Russia.
The data included in my analysis of voting unity, then, are not drawn
from a random sample of legislatures in the sense of being generated by
some stochastic process from among all democratic regimes worldwide.
Rather, the sample of national legislatures examined in this chapter and the
1
For this, legislative scholars owe gratitude to sustained effort and a commitment to data
sharing by Professor Keith Poole of the University of California, San Diego.
2
One exception was Israel, where an Israeli colleague put me in contact with a Knesset staffer
who had access to archives and was willing to transcribe hard copy records of votes into
spreadsheet format. Meanwhile, legislative scholars with diverse language skills and insti-
tutional contacts should make it a priority to determine where any yet-untapped archives of
recorded legislative votes are, and to collect and disseminate those data. To facilitate this
effort, VoteWorld: The International Legislative Roll-Call Voting Website (http://ucdata.berkeley
.edu:7101/new_web/VoteWorld/voteworld/) is a data clearinghouse established in collabora-
tion by scholars committed to open access. All the data for this project are available either via
VoteWorld or at http://www.dartmouth.edu/~jcarey.
103
Counting Votes
next includes all those where I know data to exist, making this, to my
knowledge, the broadest cross-national study of legislative voting under-
taken to date.
The samples of votes from these legislatures are shaped by vote avail-
ability and resource constraints, such as the labor to transfer each vote to
machine-readable format. Some assemblies post vote records on line as lists
of names that, with some work, can be prepared for analysis. In other cases,
my travel to the assemblies themselves aimed at locating and collecting
hard copies of whatever votes were recorded, or else contracting with local
assistants to collect the information. Scholars who had collected vote data
in similarly painstaking fashion were, by and large, generous in making
those available. In some cases (e.g., United States, Uruguay), my sample
of votes represents all recorded votes during complete legislatures; in
others (e.g., Chile, Israel, Peru), it includes all votes between specified
dates; and in still others (e.g., Canada, Czech Republic, Russia) the sample
includes those votes that colleagues made available.
What are the implications for how we interpret measures of voting
unity? If unity is fundamentally different in legislatures where recorded
votes are unavailable, then the measures reported here may be unrepresent-
ative of the whole population of legislatures. If the factors correlated with
higher and lower levels of unity are different in legislatures where recorded
votes are unavailable, then the explanations offered here for levels of voting
unity are limited to environments where votes are visible. Time spent ob-
serving legislative behavior in legislatures both where votes are visible and
where they are not leads me to only a modest conjecture on the possibility of
differences in the power of presidential patronage to sway votes in visible-
versus invisible-vote systems. I discuss these in the following chapter. That
said, it must be noted that where recorded votes are unavailable, we cannot
know for sure what levels of voting unity are and what explains them.
I draw on recorded vote data from lower legislative chambers across
nineteen countries. The unit of analysis is the party group during a given
legislature. I calculated voting indices for each party in each legislature that
it enjoyed representation (%WONi and WUNITYi), or where its group
consisted of two or more legislators (WRICEi and ULOSERi), or three or
more legislators (RLOSERi). The dates, the total number of votes, and
properties of the corresponding mean CLOSEj scores for the cases exam-
ined in this chapter are shown in Table 5.1.3
3
For an explanation of CLOSEj scores, see Appendix 2.
104
Data on Recorded Votes
Table 5.1. Recorded Vote Data
Country Dates of Assemblies No. of Votes Sum Mean

CLOSE CLOSE
Argentina December 1984– 20 12 0.62
December 1986
December 1987– 20 14 0.71
September 1989
December 1989– 65 39 0.6
December 1991
December 1991– 27 14 0.53
December 1993
December 1993– 64 35 0.55
December 1995
December 1995– 21 16 0.77
November 1997
Australia May 1996–July 1998 457 308 0.67
Brazil January 1989– 57 33 0.57
December 1990
March 1991–January 166 104 0.63
1995
March 1995– 452 291 0.64
December 1998
Canada May 1994–April 1997 735 398 0.54
Chile May 1997–January 215 59 0.27
1998
October 1998–May 522 167 0.32
2000
Czech January 1993–June 5,067 2,149 0.42
Republic 1996
July 1996 – December 4,741 2,075 0.44
1998
Ecuador July 1998–June 2002 22 5 0.25
France IV July 1946–June 1951 365 175 0.48
Republic
July 1951–June 1956 352 246 0.7
June 1956–June 1958 172 109 0.63
Guatemala December 1994– 10 6 0.58
November 1995
February 1996–January 42 21 0.51
1999
January 1999–April 2000 7 5 0.75
(continued)
105
Counting Votes
Table 5.1 (continued)
Country Dates of Assemblies No. of Votes Sum Mean

CLOSE CLOSE
Israel October 1999–November 598 205 0.34
2000
Mexico October 1998–April 2000 299 113 0.38
New Zealand November 1990–August 592 384 0.65
1993
December 1993– 185 145 0.78
November 1994
Nicaragua January 2000–September 693 417 0.62
2000
Peru March 1999–June 2000 689 430 0.33
August 2000–December 332 227 0.26
2000
August 2001–October 103 129 0.09
2001
Philippines July 1995–April 1997 147 3 0.02
Poland October 1997–May 3,045 1,226 0.4
1999
Russia January 1996–May 356 197 0.55
1997
United January 1991–December 901 495 0.55
States 1992
January 1993–December 1,094 666 0.61
1994
1996
1998
Uruguay October 1985–November 41 28 0.68
1989
December 1990–August 22 10 0.47
1994
There is tremendous variance across the legislatures for which I have

voting data in how many votes are recorded and thus available for analysis
of party unity. There is also variance in what information is available about
each vote (e.g., origin of the initiative, issue area, whether final passage).
The only information available for every vote in every chamber is
date, threshold for approval, and how each member of the assembly voted
106
Describing Voting Unity
(e.g., aye, nay, abstain, or no vote). Finally, there is variance in the overall
tendency toward consensus or contestation in votes. Mean CLOSEj sum-
marizes the extent to which an average vote was contested for each case.
Votes were most narrowly won in New Zealand, Argentina, the French
Fourth Republic, and Guatemala, less so in Ecuador, Chile, Peru, and
especially the Philippines. In all legislatures, some votes are consensual,
but in most there are deep divisions on many votes as well – enough that we
can be confident that the real fights over policy have not all ended before
votes come to the floor.
5.5. Describing Voting Unity
5.5.1. Cross-National Patterns

I calculated the various voting unity indices described in this chapter for each
party within every legislative period from which I obtained data. Across all
the periods in all the legislatures, this amounts to more than 300 observations
on each of the indices.4 Each index, in turn, summarizes a large amount of
information about individual legislators’ actions, and these descriptive sta-
tistics could be summarized and examined in myriad ways.
Table 5.2 presents the four main indices aggregated at the level of
country, along with the standard deviation across parties within each
country.5 The highest average WRICEi is found in Australia and New
Zealand while the lowest are in Nicaragua, Poland, and Russia. Recall,
however, that the extremely low indices for Nicaragua and Russia reflect
the decision to treat nonvotes as nay votes in these legislatures. When
nonvotes are discarded in calculating WRICEi, the indices shoot up in
both cases.
WUNITYi does not track WRICEi perfectly. The two are correlated at
.55 across all parties, and although Australia is highest on WUNITYi as
well, the French Fourth Republic, the United States, and Uruguay are close
behind. Meanwhile, along with Nicaragua and Russia (again, owing largely
to the coding decision), Canada, Israel, Peru, the Philippines, and Poland
4
There are the fewest observations on RLOSERi, because it is not calculated for parties with
fewer than three members. There are more observations on the other indices.
5
Table 5.2 summarizes information on four voting unity indices across all the countries
examined. It is useful as a reference source, but the format does not facilitate comparison
across cases. Figures 6.2 and 6.3 illustrate more clearly the patterns of WRICEi and RLO-
SERi and their variance cross-nationally.
107
Counting Votes
Table 5.2. Voting Unity Index Averages and Standard Deviations by Country
WRICE WUNITY RLOSER ULOSER
Country Mean s.d. Mean s.d. Mean s.d. Mean s.d.

Argentina 0.86 0.17 0.51 0.21 0.003 0.011 0.020 0.095
Australia 0.99 0.02 0.69 0.26 0.000 0.000 0.000 0.000
Brazil 0.75 0.17 0.59 0.15 0.008 0.019 0.014 0.025
Canada 0.82 0.25 0.42 0.24 0.001 0.002 0.002 0.003
Chile 0.82 0.15 0.48 0.12 0.003 0.006 0.022 0.020
Czech Republic 0.87 0.08 0.53 0.16 0.006 0.008 0.049 0.037
Ecuador 0.92 0.09 0.71 0.16 0.006 0.016 0.006 0.016
France 0.85 0.14 0.68 0.18 0.011 0.015 0.021 0.020
Guatemala 0.83 0.21 0.65 0.22 0.000 0.000 0.001 0.005
Israel 0.88 0.23 0.44 0.14 0.002 0.002 0.048 0.040
Mexico 0.84 0.16 0.63 0.21 0.011 0.010 0.025 0.015
New Zealand 0.96 0.02 0.59 0.21 0.018 0.022 0.108 0.239
Nicaragua 0.36 0.1 0.36 0.10 0.038 0.023 0.038 0.023
Nicaragua NV~ ¼ nay 0.95 0.01 0.6 0.04 0.038 0.023 0.057 0.030
Peru 0.8 0.14 0.45 0.14 0.006 0.008 0.013 0.017
Philippines 0.70 0.28 0.45 0.14 0.000 0.000 0.000 0.000
Poland 0.42 0.2 0.35 0.17 0.026 0.027 0.027 0.027
Russia 0.55 0.14 0.55 0.14 0.010 0.009 0.132 0.045
Russia NV~ ¼ nay 0.94 0.05 0.66 0.16 0.010 0.009 0.018 0.012
United States 0.70 0.06 0.68 0.06 0.119 0.038 0.133 0.045
Uruguay 0.79 0.25 0.67 0.25 0.037 0.034 0.037 0.034
all exhibit low values. Parties in the United States average the highest rates
of losses because of both cross-voting (RLOSER) and failure to mobilize
fully (ULOSER), although relatively high rates on RLOSER are also pres-
ent in Nicaragua, Poland, and Uruguay, and on the latter in various coun-
tries, although the preceding discussion of nonvoting equilibria suggests
wariness toward that statistic. A few countries, including Australia again
plus Canada, Guatemala, and the Philippines, show no – or almost no –
vote losses due to breaches in voting unity.
5.5.2. Looking at Legislatures

The next chapter focuses on explaining levels of voting unity according to
these various measures. Before closing, it will be useful merely to demon-
strate, with a few more graphs like those already presented, that the indices
allow us to visualize legislatures in ways that illustrate key characteristics of
108
Figure 5.5 WRICE, WUNITY, RLOSER, and ULOSER indices for Australia,
1996–98.
their party systems. The statistics presented in Table 5.2, for example,
suggest Australia as the prototype of a highly unified legislative party sys-
tem. Figure 5.5 illustrates this uniformity by juxtaposing its WRICEi,
WUNITYi, RLOSERi, and ULOSERi indices during the 1996–98 period.
Again, government parties are white and opposition shaded various gray
hues. The relative simplicity of Australia’s coalition structure during this
period and the regularity of legislative voting behavior are clear.
Compare this with the structure of the Israeli Knesset, shown in Figure
5.6, during a period from October 1999 to November 2000 from which I
collected a sample of votes. First, the far greater fragmentation of the
Israeli party system and governing coalition structure is evident. Israeli
parties also show somewhat more variance in voting unity across the
different indicators, and their overall levels demonstrate less unity and
some vote losses, even among parties in the governing coalition, because
of disunity.
109
Counting Votes
Figure 5.6 WRICE, WUNITY, RLOSER, and ULOSER indices for Israel,
1999–2000.
A similar complexity and fragmentation are also evident in Brazil’s party

system during the first administration of President Fernando Henrique
Cardoso, 1995–99, shown in Figure 5.7. Here, the president’s party is
shown, toward the right of each panel, in black, with other parties in the
governing coalition (i.e., holding cabinet portfolios) in white, and the var-
ious others in gray hues. As scholarship on that country frequently notes,
Brazil exhibits relatively low voting unity overall, by any index, and high
levels of vote losses because of disunity. But there is also substantial vari-
ance across parties. The Workers’ Party (PT) and the Communist Party of
Brazil (PC do B), both in opposition, for example, show high levels of unity
and mobilization, and virtually no losses owing to disunity. The governing
parties, by and large, are less unified.
Finally, consider the United States House of Representatives during the
104th Congress, 1995–97, in Figure 5.8. The two-party hegemony of the
Republicans and Democrats stands in sharp contrast to the fragmented
110
Figure 5.7 WRICE, WUNITY, RLOSER, and ULOSER indices for Brazil,
1995–1998.
Israeli and Brazilian systems. The relatively low WRICEi indices illustrate
a substantial amount of cross-voting with which even casual observers of
the U.S. Congress are familiar, but the near parity between WRICEi and
WUNITYi also shows that U.S. legislators are diligent about getting to the
floor to vote. Most of the disunity in the U.S. Congress comes in the form
of cross-voting, rather than nonvoting. As a result, while both the loss-
based indices are high for the United States, RLOSER is nearly as high as
ULOSER, especially for the Republicans, the majority party during this
Congress.
Because the voting unity indices developed in this chapter summarize
vast amounts of legislative activity in relatively simple statistics, it possible
to compare distinct legislatures, or the same legislature in different periods,
or both, according to common metrics. This is useful for describing and
visualizing legislative party systems, but the goal of developing these tools
111
Counting Votes
Figure 5.8 WRICE, WUNITY, RLOSER, and ULOSER indices for the
United States, 1995–97.
is to explain the conditions that generate high and low levels of voting
unity, mobilization, and ultimately partisan and government success or
failure on the floor. The next chapter focuses on that task.
Chapter 5 Appendix 1: Illustrations of RICE and UNITY Indices

Appendix Table 5.1 shows examples of RICE and UNITY scores, and their
relative sensitivity to both cross-voting and nonvoting, in a hypothetical
300-member legislature with three parties of 100 members each. The first
column shows the overall tally for each vote in the format (aye, nay, non-
vote). The next column shows values for a measure of how closely contested
each vote is. The next columns show the tally for each party on each of six
votes, followed by the parties’ UNITYij and RICEij scores.
On the first vote, all three parties mobilize only half their members, but
within each party all members who cast decisive votes vote the same way;
thus UNITYij scores in each case are relatively low at .50, whereas RICEij
112
Appendix 1
scores are ‘‘perfect’’ at 1.00. Voting participation increases across the six
votes in every party – that is, the number of nonvotes declines. Now consider
the pattern of Party A, which experiences increasing cross-voting as more of
its members cast decisive votes; thus, its RICEAj scores plummet faster than
its UNITYAj scores, converging at zero on the last vote. Party B experiences
substantial cross-voting on early votes but pulls together subsequently such
that both RICEBj and UNITYBj rise and converge as more legislators mo-
bilize on later votes. Party C experiences no cross-voting on any votes, so
UNITYCj rises to converge with RICECj as mobilization increases.
Chapter 5 Appendix 2: Weighting by CLOSEness

When votes are consensual in the legislature as a whole, voting unity scores
for any subset of legislators will necessarily be high, at least as measured by
RICEij. Counting all votes equally, including lopsided ones, therefore,
inflates unity indices. This presents a particular problem for cross-national
comparisons where there is variance across cases in the average closeness of
votes owing to characteristics of legislatures entirely unrelated to party
unity. For example, if rules in Legislature A require recording votes on
every motion, the vast majority of which are perfunctory and consensual,
whereas in Legislature B only votes on substantive (and potentially divisive)
motions are recorded, then unweighted indices from the two legislatures
would show higher unity in A, even in the absence of any real effect on
legislative decision making.
The conventional response in studies of recorded votes is to establish
some criterion for throwing out votes that are ‘‘too consensual’’ to
be considered relevant to party unity. Established criteria in studies of
the two-party U.S. Congress often focus on whether the majorities or
the leaderships of the two main parties oppose each other on a given vote
(Brady, Cooper, and Hurley 1977; Cox and McCubbins 1993). In the
multiparty environment of most other democracies, however, such criteria
are of little use. Which votes meet the selection criterion would vary
according to which parties’ majorities or leaderships are considered. Ano-
ther approach is to include all votes on which some minimum proportion of
legislators votes on the losing side (Mainwaring and Pérez-Liñán 1997;
Figueiredo and Limongi 2000). But such thresholds are necessarily arbi-
trary, and they count all votes, no matter how far above threshold, equally,
contradicting the basic intuition behind selection criteria in general – that
the sternest test of unity is whether members of a party or coalition vote
113
114
Appendix Table 5.1. Examples of RICE and UNITY Scores, and Weighted and Unweighted Indices, in a Hypothetical Legislature
Party A Party B Party C
Tally CLOSEj TallyA UNITYAj RICEAj TallyB UNITYBj RICEBj TallyC UNITYCj RICECj
[150,0,150] 0.00 [50,0,50] .50 1.0 [50,0,50] .50 1.00 [50,0,50] .5 1.00
[162,18,120] .20 [54,6,40] .48 .80 [48,12,40] .36 .60 [60,0,40] .6 1.00
[168,42,90] .40 [56,14,30] .42 .60 [42,28,30] .14 .20 [70,0,30] .7 1.00
[168,72,60] .60 [56,24,20] .32 .40 [32,48,20] .16 .20 [80,0,20] .8 1.00
[162,108,30] .80 [54,36,10] .18 .20 [18,72,10] .54 .60 [90,0,10] .9 1.00
[150,150,0] 1.00 [50,50,0] 0.00 0.00 [0,100,0] 1.00 1.00 [100,0,0] 1.0 1.00
Unweighted
indices .32 .50 .45 .60 .75 1.00
Weighted
indices .20 .27 .55 .60 .87 1.00
Appendix 2
together when doing so matters to legislative outcomes, and therefore

that the more hotly contested a vote is, the more relevant it is to a measure
of unity.
This suggests RICEi and UNITYi indices calculated as follows:
X X
WRICEi ¼ RICEij CLOSEj = CLOSEj
X X
WUNITY i ¼ UNITY ij CLOSEj = CLOSEj ;
where
CLOSEj ¼ 1 ð1=THRESHOLD jTHRESHOLD %AYEjÞ6
for legislature as a whole on vote j.

These indices are summary statistics for voting unity in party i, weight-
ing votes by how closely they were contested, according to the basic in-
tuition that, for a party seeking to influence outcomes, unity is more critical
the more likely it is that defection (or defection and nonvoting, in the case
of UNITYi) of any member(s) will be pivotal.
In Appendix Table 5.1, the bottom two rows illustrate this weighting
system for the hypothetical legislature. The second column of the table
shows how closely contested each vote was (CLOSEj is described in the
appendix). On each successive vote in this example, the legislature as
a whole is more closely divided, and these more closely contested votes
count more heavily under the weighting scheme than if the index were
a simple, unweighted mean of scores.
Because Party A displays increasing cross-voting on the more conten-
tious votes, and because cross-voting drives down UNITYAj and RICEAj
alike, Party A’s weighted indices are both well below its corresponding
unweighted indices. Next, consider Party B. The first vote is consensual,
so all parties are, by definition, perfectly unified as measured by RICE
scores (although their UNITY scores lag because of nonvoting). Beyond
this vote, Party B exhibits greater cross-voting on the relatively lopsided
votes and pulls together on the more closely contested and heavily
weighted ones. Because more closely contested votes count more under
6
When the threshold for passing a measure is a simple majority of those voting, the formula
can be written as: CLOSE ¼ 1 (2 * |50% %AYE|). However, when passage requires an
extraordinary majority, the more general formula still applies. This form of the general
equation was suggested to me by Jeanne Giraldo.
115
Counting Votes
the weighting system, WUNITYB is higher than UUNITYB, and WRICEB

pulls even with URICEB. Party C exhibits a similar pattern, with no cross-
voting and so perfect RICE scores but increasing mobilization on closer
votes. The trajectories of UNITY and RICE scores for each party as close-
ness varies rise across the six votes described in Appendix Table 5.1, as are
illustrated in Appendix Figure 5.1. By and large, if a voting unity index rises
with closeness, then its weighted measure > unweighted, whereas if it is
inversely related to closeness, the opposite is true.
Chapter 5 Appendix 3: Legislative Success Index

To determine whether a party ‘‘won’’ on a given measure, it is necessary to
infer each party’s preference. I rely on the votes themselves, attributing to
each party the preference supported by the majority of its voting members.
For every party, i, on every vote, j:
PREFij ¼ Approve if AYEij > NAY ij , Reject if NAY ij < AYEij ,

No Preference AYEij ¼ NAY ij
Thus if most of a party’s decisive votes were aye and the measure is
approved, it counts as a win; if most of its votes were aye and the measure
is rejected, it counts as a loss, and so on.7
Chapter 5 Appendix 4: RLOSER and ULOSER

RLOSERij ¼ 1 if:
PREFij ¼ Approve, AND Outcomej ¼ Reject, AND TotalAYEj þ
NAYij > Thresholdj,
or if
PREFij ¼ Reject, AND Outcome j ¼ Approve, AND TotalAYEj –
AYEij < Thresholdj
ULOSERij ¼ 1 if:
PREFij ¼ Approve, AND Outcomej ¼ Reject, AND TotalAYEj þ
NAYij þ NONVOTESij > Thresholdj,
7
In a handful of legislatures, such as Brazil’s Chamber of Deputies and the U.S. Congress on
some votes, parties’ formal positions on specific measures are reported as part of the assem-
bly’s published record. This practice, however, is sufficiently rare as not to be viable for
broad cross-national analysis.
116
Appendix Figure 5.1 Patterns of UNITY and RICE scores, and effects on
weighted versus unweighted indices, as CLOSEness varies.
117
Counting Votes
or if
PREFij ¼ Reject, AND Outcomej ¼ Approve, AND TotalAYEj
AYEij – NONVOTESij < Thresholdj
where, for every vote, j:
Thresholdj ¼ number of votes necessary to approve the measure
Outcomej ¼ [Approve, Reject]
RLOSERij and ULOSERij identify votes on which, given how all other
parties voted, a party lost despite the fact that it could have prevailed had it
been fully unified or mobilized. Appendix Table 5.2 illustrates some sce-
narios in a hypothetical legislature with 100 members and two parties, A
and B, with 60 and 40 seats, respectively.
A party ‘‘loses’’ a vote whenever the outcome runs contrary to that
supported by a majority of its voting members. When the parties vote
together, neither loses, as in Votes 1 and 2. When both parties vote along
party lines (high RICEij scores for both), as in Votes 3 and 4, Party B loses
provided that UNITYij scores are closely correlated – that is, provided that
both parties mobilize around the same proportion of their members to
vote. Note, however, that on Vote 4, Party B could have prevailed, given
how Party A voted, had it mobilized its full complement of legislators to
vote nay. Thus, on Vote 4, Party B is a ULOSER.
Votes 5 through 8 represent losses by Party A because of disunity of
various sorts. Vote 5 shows a straightforward breakdown within Party A,
with some members voting against the party majority, swinging the out-
come in favor of the united and mobilized Party B. The results of Vote 6
are similar, where Party A suffers a combination of defections among
voting and nonvoting members while confronting a unified and mobi-
lized opposition. In Votes 7 and 8, Party A is both divided and fails to
mobilize, while Party B lags on either one or the other count but prevails
on the vote outcome. In each of these cases, Party A’s disunity allows B to
win (i.e., costs A the vote). On Votes 5, 6, and 7, Party A could have won
(i.e., passed the measure) if all of its voting members had voted aye, and
obviously had it fully mobilized. In these cases, therefore, Party A is both
the RLOSER and ULOSER. On Vote 8, Party A could have won had it
fully mobilized, but not merely had all its voting members voted aye, so it
is a ULOSER but not an RLOSER.
As long as UNITYij is strongly correlated across the parties on a given
vote, the outcome will reflect the distribution of seats across parties.
118
Appendix Table 5.2. Illustrations of RLOSER and ULOSER in a Hypothetical 100-Member Legislature (tallies are [aye-nay-nonvote])
Vote TALLYAj TALLYBj Losing RICEAj RICEBj RLOSER UNITYAj UNITYBj ULOSER
Party
1 [60–0–0] [40–0–0] None 1.00 1.00 None 1.00 1.00 None
2 [30–0–30] [20–0–20] None 1.00 1.00 None .50 .50 None
3 [60–0–0] [0–40–0] B 1.00 1.00 None 1.00 1.00 None
4 [30–0–30] [0–20–20] B 1.00 1.00 None .50 .50 Party B
5 [45–15–0] [0–40–0] A .50 1.00 Party A .50 1.00 Party A
6 [40–10–10] [0–40–0] A .60 1.00 Party A .50 1.00 Party A
7 [20–10–30] [0–20–20] A .33 1.00 Party A .17 .50 Party A
8 [15–5–40] [10–30–0] A .50 .50 None .17 .50 Party A
119
Counting Votes
Outcomes unreflective of the seat distribution become possible when UN-

ITYij scores come uncoupled. In the example of Appendix Table 5.2, where
there is a majority party, for Party A to be defeated, its UNITYAj must drop
more than that of its opponents. The same does not necessarily apply for
RICEij scores, as Vote 8 shows.
Given that RLOSERij and ULOSERij are calculated with respect to the
outcome of each vote, I do not weight them in creating summary indices,
RLOSERi and ULOSERi, for each party, but simply report proportions –
for example, on how many votes, out of all votes analyzed, was a party an
RLOSER.
Chapter 5 Appendix 5: Nonvotes and Relative versus Absolute-Vote

Thresholds
Consider first the mechanics of the rules for approving measures that are
put to a legislative vote. Although some votes may require extraordinary
majority support (60%, 67%, or 75%) for approval, most votes in most
legislatures require simple majority support (> 50%). But a majority of
whom? In most cases, chamber rules stipulate the proportion of members
that constitutes a quorum, and approval of standard measures requires
support from a majority of those voting aye or nay when a quorum is
present. Thus, the precise number for approving a measure is set in relative
terms – relative to the number voting.
Under a relative threshold, if I disagree with my party’s position, either I
might withhold my support from the party by not voting (whether through
abstention, or not showing up, or simply not pressing a button on my
electronic voting device), or I could not only withhold my support but also
give my vote to the other side. The latter is a more visible breach of unity
than the former and does correspondingly more damage to the party’s
collective brand name; and if the vote is close, the latter is twice as dam-
aging to my party’s prospects of winning than the former. This difference is
at the heart of the distinction between the traditional RICEij score and the
UNITYij score developed previously.
In some legislatures, however, thresholds are set in absolute terms, as a
percentage of the full membership of the assembly. Among the cases analyzed
in this book, the Russian Duma and Nicaraguan and Guatemalan assemblies
require absolute thresholds to pass any measure. Under such rules, nonvotes,
whatever their intent, are equivalent to nays in their effect on outcomes. For
the purposes of calculating voting unity indices, my point of departure is to
120
Appendix 5
Appendix Figure 5.2 Nicaraguan WRICE indices.
treat them as such – that is, to count nonvotes as nays when their effects on
outcomes are equivalent to nay votes.
This approach warrants careful consideration, however, because count-
ing nonvotes as nays renders RICEij scores, in particular, and the indices
built from them, highly sensitive to nonvoting. For example, Appendix
Figure 5.2 compares WRICEi indices for parties in the Nicaraguan
121
Counting Votes
Assembly, first calculated with nonvotes as nays, then discounting nonvotes

altogether. In the figure, weighted RICEi indices are represented by the
height of each bar on the y axis. Each party’s share of assembly seats is
represented by the width of its column along the x axis.8
The second panel, with near-perfect WRICEi indices across the board,
illustrates that cross-voting was nearly absent from this Nicaraguan Assem-
bly. Yet ignoring nonvotes, as in this panel, overestimates party voting
unity as it affects vote outcomes, because any legislator who does not like
a measure her party supports can oppose it as effectively by not voting as by
overtly crossing the aisle to vote nay. As the WRICEi indices in the first
panel show, Nicaraguan deputies did not reliably deliver their votes to
support their parties’ positions.
How, then, ought one treat nonvoting under absolute-threshold voting
rules? Cautiously, and with explicit consideration of what one is looking for.
With respect to outcomes, nonvotes are equivalent to nays under absolute
thresholds, so if one is interested in effects of votes on outcomes, nonvotes ¼
nays is appropriate. With respect to the communicative element of voting –
the extent to which legislative votes are expressions of a party’s policy posi-
tions – the situation is murkier. Nonvoting is, effectively, passive opposition
to a measure. If passive opposition is markedly less visible to copartisans and
to citizens than explicit opposition, then not voting when one’s party sup-
ports a measure represents something less than a full-scale breach of party
unity, but ignoring nonvotes altogether probably fails to capture the dissent
they imply. Moreover, when one’s party opposes a measure, ignoring non-
votes fails to capture the party unity they entail.
Chapter 5 Appendix 6: Measuring Voting Unity in Small Parties

The various measures of voting unity confront three types of limitations
associated with small parties. First, the RICEij score is not relevant for
a party with only one member because cross-voting is, by definition, im-
possible. Thus, RICEij scores are not calculated for parties with only one
legislator, or for votes on which only one member of a party participates.
Second, RLOSERij is calculated only for parties where N > 2. RLOS-
ERij is derived from simulated vote outcomes under alternative, ‘‘more
unified’’ permutations of a party’s votes, given the party’s inferred
8
The columns do not necessarily reach the 100 percent mark, as indices are not calculated for
independents who remained unaffiliated with any party bloc or for single-member parties.
122
Appendix 6
preference on the vote. Where N 2, the party either has no inferred

preference (i.e., splits [1–1]) or is perfectly unified, in which case no alter-
native, more unified permutation is possible. ULOSERij can be calculated
for parties with two members or more because, for example, a [1–0–1] tally
indicates a partisan preference for aye, on which the party could have
mobilized more effectively with [2–0–0].
The third consideration is that both RICEi and UNITYi are subject to
upward bias as a combined function of a group’s size and the underlying
proclivity of its members to vote together (Desposato 2005). The bias is
more severe the smaller the group and the less inclined its members are to
vote alike. The problem is that the probability of observing instances of
high party unity (e.g., all voting aye, or all nay) is higher the fatter are the
tails of the binomial distribution of the proportion of ‘‘alike’’ votes. Obser-
vations in these tails reflect higher RICEi and UNITYi values than the
underlying probability of voting alike would suggest, biasing the measures
upward. The tails of the distribution are fatter when N is smaller, and the
resulting bias is more pronounced when the underlying probability of
voting together is smaller. (Think of the likelihood of observing all ‘‘heads’’
– that is, ‘‘perfect unity’’ – when tossing a pair of coins, as opposed to when
tossing ten coins.) The magnitude of small-group bias declines rapidly as
party size and underlying cohesiveness increase.
Desposatos’s (2005) analysis suggests that the potential bias in cohesive-
ness scores can be corrected by estimating deviance factors for RICE and
UNITY scores, which are functions of group size and the underlying pro-
clivity of members to vote together, then subtracting that factor from the
score. The process I use is as follows. For any party i on vote j, one can
calculate the proportions of legislators who vote together (T) with most of
the group, or who vote in dissent (D):
Tij ¼ maximum [AYE, NAY], as a percentage of those voting
Dij ¼ minimum [AYE, NAY], as a percentage of those voting
One can also calculate analogous proportions of legislators mobilized
(M), those opposed (O), and those not voting (NV) based on the size of the
group, rather than just those who vote:
Mij ¼ maximum [aye, nay] as a share of all legislators
Oij ¼ minimum [aye, nay] as a share of all legislators
NVij ¼ 1 - Mij – Oij
The RICEij score is just Tij – Dij, and the UNITYij score is Mij – Oij.
123
Counting Votes
The corresponding RICEi and UNITYi indices are summations of

Ti – Di, and Mi – Oi across all votes. The indices, then, reflect estimates
of the underlying probabilities of voting ‘‘against’’ the group, or withhold-
ing one’s vote from the group.
For each party group, i, I calculate the expected upward bias due to small
party size as:
RICEdeviancei ¼ Di / Ni
UNITYdeviancei ¼ Oi / Ni
where Ni is the number of members in the cohort. I then calculate the
‘‘empirically corrected’’ indices for each cohort by subtracting its deviance
factor from its ‘‘raw’’ index. The indices are ‘‘empirically corrected’’
because the estimates of underlying probabilities of Di and Oi are based on
the observations of behavior across all votes. The deviance factors grow as
the probability of cross-voting grows, and shrink as Ni grows. For exposi-
tional simplicity, I do not include the word ‘‘corrected’’ each time I refer to
the corrected indices, but all indices presented in this book are corrected
for potential bias.
124
6
Explaining Voting Unity
6.1. Legislative Parties and Institutional Context

The institutional environment in which parties operate is widely held to
affect their voting unity. Parties in parliamentary systems are generally
characterized as highly unified, and those in presidential systems as more
fractious and less disciplined, with resulting difficulty for presidents in the
legislative arena (Diermeier and Feddersen 1998; Hix, Noury, and Roland
2006; Persson and Tabellini 2003; Shugart 1998). Federalism, by encour-
aging the organization of parties at the subnational level, is posited to foster
divisions within parties at the national level (Mainwaring 1999; Weyland
1996). Electoral systems that provide for competition among legislative
candidates within the same party for personal votes are portrayed as encour-
aging disunity relative to closed-lists election rules (Ames 1995; Golden and
Chang 2001; Hix 2004). The leadership of parties that are older and better
established may be more autonomous and less vulnerable to pressure from
presidents (Stokes 2001).
These assertions are not uniformly accepted. On the basis of a broad cross-
national study, Cheibub, Przeworski, and Saiegh (2004) argue that presidents
are on par with parliamentary executives in forming legislative coalitions to
pass legislation. After completing a case study of Brazil, a presidential federal
system with intraparty electoral competition – all the characteristics listed as
undermining party unity – Figueiredo and Limongi (2000) argue that various
provisions centralizing control over the legislative agenda provide leverage to
control wayward parliamentarians and govern as efficiently as governments
that confront none of these institutional obstacles ostensibly do.
It is difficult to know whether institutions matter to party unity, which
institutions, and how much, without cross-national studies with sufficient
125
breadth to allow for variance in the institutional factors of interest.

Morgenstern (2003) makes an ambitious contribution along these lines,
but his empirical analysis includes five countries, all presidential, which
limits his ability to test for the effects of constitutional structure, and its
interaction with party-level factors, on voting unity. Sieberer’s (2006) study
is similarly constrained by including only European parliaments. Drawing
on the data and measures of unity described in Chapter 5, this chapter tests
for how the institutional environment affects legislative party unity.
I begin by reviewing three distinct potential mechanisms that might
produce party unity in legislative voting – cohesiveness, discipline, and
agenda control – and argue that the logic of competing principals operates
through the first and second of these. Then I use cross-national voting data
to shed some light on the relationship between cohesiveness and discipline
across the parties included in this analysis. Drawing on the logic of com-
peting principals, I spell out a number of explicit hypotheses regarding
factors that should affect legislative voting unity. After presenting the mod-
els used to test the hypotheses against the cross-national data and the
results, I extend the analysis of voting unity from individual parties to
governing and opposition coalitions.
6.2. Competing Principals and Existing Accounts of Party Unity

There are three potential sources of legislative party unity: cohesiveness,
discipline, and agenda control. Cohesiveness implies that elections produce
legislative parties whose members have similar preferences and therefore
vote in harmony. Discipline refers to the combination of responses, gen-
erally administered by party leaders, used to reward voting loyalty and
deter or punish breaches in solidarity. Strong discipline should raise party
voting unity, other things being equal. Agenda control implies that those
who control the flow of legislative traffic steer it so as to determine whether
proposals that would divide a given party or coalition come to a vote.
Where control of a legislative chamber’s agenda is monopolized by the
leaders of a given party or coalition – an agenda cartel – we should expect
them to minimize agenda access for measures that would divide the party,
and so we might expect that levels of voting unity should vary according to
which parties control the agenda (Cox and McCubbins 2005).
The competing principals theory advanced in this book derives hypotheses
based on electoral sources of party cohesiveness and the institutional resour-
ces that drive discipline within parties. For example, electoral rules that
126
Competing Principals and Existing Accounts of Party Unity
centralize control over nominations allow party leaders to screen potential

candidates for ideological compatibility with national party priorities, and so
to foster cohesiveness among those ultimately elected. The same centraliza-
tion over nominations – or of list positions in closed-list systems – also
provides leaders with formidable sanctions should legislators breach party
voting unity against the leadership’s wishes. Mavericks can be denied re-
nomination or moved down lists to marginal or unwinnable positions.
It is easy to posit how other institutional factors might also affect party
cohesiveness or discipline. For example, parties in federal systems are more
apt to have autonomous subnational organizations than those in unitary
systems. If the same party’s reputation and priorities vary according to
diverse interests across subnational electorates, then national legislative
party groups should be less cohesive in federal than in unitary systems.
As another example, presidentialism creates a potential rival to legislative
party leaders, endowed with considerable resources to command respon-
siveness among legislators to an alternative set of demands, possibly under-
mining the effect of legislative party discipline.
The competing principals explanation, therefore, bears directly on both
cohesiveness-based and discipline-based stories for why party unity may be
high or low. A competing principals theory of party unity, by itself, is less
directly relevant to the agenda cartel account of party unity, for a number of
reasons. Agenda cartel theory focuses on the direction of policy changes
under specific governing coalitions, whereas the competing principals
model does not make claims about legislative outcomes in a theoretical
policy space (Cox and McCubbins 2005). The implications of agenda con-
trol theory regarding voting unity focus on the specific subset of legislative
votes – on key procedural matters or those that ratify final passage of new
policies, for example. Unfortunately, the data available across the range of
legislatures included in this analysis often do not make it possible to iden-
tify and separate out such votes. In part this is due to the quality of the data,
but in part it has to do with the diversity of legislative procedures cross-
nationally. The mechanics of final passage itself are context-specific. The
voting process whereby various alternative proposals are sifted until a
surviving contender is pitted against the status quo is characteristic of the
specific amendment procedure used in the U.S. Congress, but not in many
other legislatures (Rasch 2000; Weldon 1997). More broadly still, the set of
rules governing who can bring motions to the assembly floor for a vote
varies across countries and legislative chambers, with control vested in
chamber directorates in some cases, monopolized by executives drawn
127
from the chamber in others, and shared with independently elected exec-
utives in others.
In short, the data and theory presented here provide substantial leverage
in testing for party unity driven by cohesiveness and discipline across the
full spectrum of legislative votes, but more limited leverage in testing for
unity driven by the specific mechanism and conditions posited by agenda
cartel theory. The results here shed light on cohesiveness of preferences
among copartisans and the relative monopoly on discipline imposed by
legislative party leaders.
6.3. Cohesiveness and Discipline: Weighted and

Unweighted Indices
The terms ‘‘cohesiveness’’ and ‘‘discipline’’ are both frequently used in
reference to the voting unity within legislative parties, but it is important
to keep in mind the conceptual distinction between the terms. The former
refers to the degree to which the members of a group share similar pref-
erences; the latter, to the degree to which group leaders are able to elicit
unified voting on the part of the group, regardless of member preferences.
Unless there is reason to believe that a particular pattern of voting behavior
is caused by either cohesiveness or discipline, I use the more generic term
‘‘unity’’ to describe the proclivity of copartisan legislators to vote together.
The difficulty of distinguishing party cohesiveness from discipline is
familiar to students of American politics, and debate over the relative con-
tribution of each to party unity in U.S. congressional voting constitutes
a substantial literature in its own right (Krehbiel 2000; Snyder and
Groseclose 2000; Cox and Poole 2004). As the previous chapter described,
I rely on a weighted index of RICEij scores (WRICEi) as one of my sum-
mary indicators of party unity, but comparing the unweighted mean
(URICEi) of a party’s RICEij scores with WRICEi can provide leverage
on whether it is cohesiveness among copartisans or discipline that accounts
for the levels of unity we observe.
Consider first the ‘‘discipline-free’’ scenario. On votes that are consensual
across an entire legislature, RICEij scores will necessarily be high for all
parties. As votes diverge from consensus, low party unity scores become
possible. Disunity within parties is still not inevitable, because lack of con-
sensus at the level of the legislature could be the product of conflict between
or among highly unified parties, but the presence of dissenting votes at the
assembly level allows for the prospect of internal party disunity whereas
128
Weighted and Unweighted Indices
consensus at the assembly level does not. The more hotly votes are contested
in the legislature overall, the more ‘‘room’’ there is, arithmetically, for dis-
unity within parties. Thus, in the discipline-free scenario, RICEij scores
decline as CLOSEj rises, and WRICEi is lower than URICEi.
Now consider the scenario with party discipline – that is, where party
leaders are able to compel their legislators to vote together. RICEij scores
must still be high on consensual votes, by definition. Where votes are
moderately contested, there is the potential for disunity within parties.
But as votes approach toss-ups (i.e., as CLOSEj approaches 1.0), such that
prevote head counts suggest a handful of switched votes one way or the
other could turn the outcome, party leaders should be increasingly inclined
to impose discipline (King and Zeckhauser 2003). Thus, the more that
discipline, as opposed to cohesiveness, accounts for levels of unity, the
more we should observe elevated RICEij scores as CLOSEj approaches
1.0 – that is, on the votes that enter most heavily in my weighting scheme.
It follows that the more that discipline, as opposed to cohesiveness, drives
voting unity, the higher the ratio of WRICEi to URICEi.1
In Table 6.1 and Figure 6.1, Party A illustrates the discipline-free sce-
nario, and Party B the disciplined scenario, for RICE scores and indices
across six votes in a hypothetical 300-member legislature. In Party A,
RICEAj declines as successive votes are contested more closely at the level
of the legislature as a whole; thus, the weighted index is substantially lower
than the unweighted. Party B experiences divisions on some moderately
contested votes but pulls together on the closest votes, with the effect that
its weighted index is higher than its unweighted. Party C experiences only
one instance of dissent, on a close vote, such that its weighted index, like
A’s, is lower than its unweighted (although not by as much).
A relevant statistic, then, is the ratio between WRICEi and URICEi.2 A
lower ratio indicates a tendency for intraparty splits, when they do occur, to
1
Snyder and Groseclose (2000) use a variation on this insight to gain leverage on the
cohesiveness-versus-discipline debate on roll call voting in the U.S. Congress. They suggest
that, whatever levels of dissent are evident, lopsided votes should nevertheless contain in-
formation about legislators’ unconstrained policy preferences (and so, about cohesiveness),
on the grounds that party leaders should have no interest in imposing discipline on votes
that are not expected to be close.
2
Note that the information in the analogous relationship between WUNITYi and
UUNITYi is, again, more ambiguous. UNITYij can be low if many members abstain or
do not vote, even on consensual votes (i.e., low CLOSEj). As a result, the relationship
between weighted and unweighted RICEi indices provides better purchase than that be-
tween weighted and unweighted UNITYi indices on cohesiveness versus discipline.
129
Table 6.1. Cohesiveness, Discipline, and Weighted and Unweighted RICE Indices in
a Hypothetical Legislature
Party A Party B Party C

Tally CLOSEj Tally RICEij Tally RICEij Tally RICEij
[300,0] 0 ¼ 1–2*|.5–1| [100,0] 1.00 [100,0] 1.00 [100,0] 1.00
[270,30] .2 ¼ 1–2*|.5–.9| [95,5] .9 [75,25] .50 [100,0] 1.00
[240,60] .4 ¼ 1–2*|.5–.8| [90,10] .8 [50,50] 0.00 [100,0] 1.00
[210,90] .6 ¼ 1–2*|.5–.7| [85,15] .7 [25,75] .50 [100,0] 1.00
[180,120] .8 ¼ 1–2*|.5–.6| [80,20] .6 [0,100] 1.00 [100,0] 1.00
[150,150] 1.0 ¼ 1–2*|.5–.5| [75,25] .50 [0,100] 1.00 [75,25] .50
WRICEi .63 .73 .83
URICEi .75 .67 .92
Weighted:Unweighted .84 1.09 .90
Figure 6.1 RICE scores for three parties: Cohesiveness versus discipline.
happen on closely contested votes. Higher ratios indicate parties that may
experience splits on lopsided votes but pull together on closer ones. Across
the cases examined here, it turns out that weighted indices tend to be
slightly lower than unweighted, but there is substantial variance. The mean
WRICE:URICE ratio ¼ .95, but the standard deviation ¼ .12. So in the
typical party, unity does not deteriorate ‘‘when it matters most’’ as
130
Weighted and Unweighted Indices
chronically as for Party A, but neither does the average party rally when the
chips are down as reliably as Party B.
If we assume that party leaders value unity more, and are less tolerant of
cross-voting, on close votes than lopsided ones, then the ratio is an indicator
of the leader’s ability to impose discipline (Snyder and Groseclose 2000;
King and Zeckhauser 2003). The ratio, therefore, provides a rough proxy
for the relative contribution of discipline to a party’s overall voting unity.
It is worth noting that ratios do not necessarily correspond to overall
levels of party unity. The following combinations are all possible:
low unity, low ratio, indicating a party that is neither cohesive nor
disciplined;
low unity, high ratio, indicating a pervasive lack of cohesiveness, but the
party pulls together more on close votes than on lopsided ones,
suggesting some measure of discipline;
high unity, low ratio, indicating a party that is generally cohesive, but
what breaches occur come on close votes, suggesting a lack of disci-
pline; or
high unity, high ratio, indicating a party that is consistently unified, and
airtight on close votes.
Comparing a party’s ratio with its level of voting unity can inform us
about the relative contributions of cohesiveness and discipline to overall
unity. In making this comparison, it is preferable to use URICEi rather
than WRICEi, the formula for which emphasizes close votes and so already
encompasses more of the information reflected in the ratio itself. Consider
Figure 6.2.3
There are cases of discipline without unity (i.e., low unity, high ratio) in
the lower right section, but the unity-without-discipline upper left corner of
the graph is sparsely populated. The ratio and URICEi are correlated at .34,
and the general pattern is that parties with higher ratios are more unified.4
3
The Philippines case is dropped because it is an extreme outlier in its near absence of even
moderately contested floor votes, leaving ratios that are hypersensitive to a handful of close
votes (see Table 5.1).
4
Plotting WRICEi against the weighted:unweighted ratio yields a tighter scatter, with no
low-ratio parties extremely low on WRICEi and a stronger positive correlation (.63), al-
though the lower right quadrant (high discipline, low unity) remains well populated. This
stronger relationship is to be expected, given that both WRICEi and the ratio are drawing
on the same proclivity of copartisans to vote together on close votes. Nevertheless, the
pattern confirms that among highly unified parties, the proclivity to vote with one’s co-
partisans is more pronounced on close votes, precisely when party leaders should be watching.
131
Figure 6.2 Scatterplots of URICEi against the ratio of WRICEi:URICEi.
The relationship is far from ironclad, and there are cases of all the combi-
nations of cohesiveness and unity outlined previously, but the overall
pattern is consistent with the idea that some measure of discipline on
close votes, in addition to innate cohesiveness, accounts for party unity
in legislative voting.
6.4. Hypotheses: Legislative Parties and Competing Principals
6.4.1. First Principals

Building on the logic of competing principals outlined in Chapter 1, I begin
with the premise that party leaders are important actors to whose demands
legislators might respond. In all democratic systems, national legislatures
are organized by parties, and almost all legislators are members of party
groups within their assemblies. To varying degrees, the leaders of these
groups control resources – appointment to key committees, control over
the legislative agenda, office space, staff, and perks – valued by rank-and-
file members. Legislative party leaders may also share command of national
party organizations, which often control resources critical to legislators’
political career prospects, such as nominations for reelection to the legis-
lature or for other offices, appointed posts, and access to campaign finance.
132
Legislative Parties and Competing Principals
Thus, virtually all legislators are subject to influence by at least one prin-
cipal – their legislative party leadership.
Whether they are subject to pressure from other, competing principals
depends on the institutional context in which they operate. The hypotheses
that follow posit the impact of various factors that affect the relative com-
mitment of legislators to the national party’s collective reputation and
those that could pull legislators in other directions.
6.4.2. Competing Principals Hypotheses

First consider the extent to which legislators’ electoral connection to voters
might pull them in directions contrary to the demands of legislative party
leaders. Where party leaders exercise strong influence over a legislator’s
election, the demands to which the legislator must respond in order to
pursue reelection and the demands from those who control the distribution
of resources within the assembly are consistent. The principal to whom the
legislator must respond on both counts is the party leadership. Where
voters exercise relatively more control over legislators’ electoral prospects
and party leaders less, legislators may face demands from their electoral
principals that compete with those of party leaders.
The rules by which legislators are elected affect their relative respon-
siveness to party leaders and to alternative interests in the electorate
(Shugart, Valdini, and Suominen 2005; Hix 2004). Where party leaders
draw up lists of candidates that are presented in general elections and
cannot be altered by voters, or can be altered only under extraordinary
circumstances, electoral responsiveness to a competing principal is mini-
mized. By contrast, where candidates compete against copartisans for voter
support and that competition determines which candidates from a party
win seats, then legislators have reason to cultivate reputations distinct from
their copartisans’.5
5
The degree to which electoral rules encourage personal vote seeking among candidates can
be parsed more precisely than the dichotomy on which I rely here. Even in Carey and
Shugart’s (1995) full rank ordering of electoral systems, however, the key distinction –
which determines whether increases in district magnitude will push toward more or less
personalism – is whether or not voters are afforded the opportunity to cast votes among
competing copartisans that determine which candidates within the party win seats. Given
the amount of empirical variation on electoral rules represented in my data, the best option
is to rely on this yes-no dichotomy on intraparty competition to characterize institutional
forces operating on the connection between legislators and their principals.
133
H1: Party unity should be lower where legislative candidates compete

against members of their own parties for personal votes than where
nominations are controlled by party leaders and electoral lists are
closed.
Next, consider the effects of unitary versus federal systems of govern-

ment. Under the former, the strongest level of party organization is gen-
erally national, the level at which the leaders who control the party group in
the national legislature operate. Under federal systems, by contrast, the
primary level of party organization, where politicians build careers and
win or lose renomination for office, is often a subnational political unit
(e.g., state or province). Heterogeneity across these units may be reflected
within parties at the national level, subjecting legislators to competing pulls
from principals at the national versus subnational levels and undermining
voting unity in the national legislature.
H2: Party unity should be lower in federal systems than in unitary ones.
The next hypotheses focus on the effects of constitutional regime type –

specifically, presidentialism versus parliamentarism – and how regime type
interacts with a party’s status in government or in opposition to affect party
voting unity. It is widely held that presidentialism undermines party unity
whereas parliamentarism fosters it (APSA 1950; Diermeier and Feddersen
1998; Gerring, Thacker, and Moreno 2005). Yet there are various ways
this might work. In this section I articulate a competing principals account
of presidential disunity, and in the next I consider an alternative account
of parliamentary unity based on the confidence vote. The key point for now
is that the implications of the accounts differ in some of the hypotheses
they suggest.
The core of the competing principals account of presidential disunity is
that popularly elected presidents, whether in pure presidential or hybrid
regimes, are potentially powerful competitors with party leaders for legis-
lators’ attention and responsiveness. Presidential elections allow the pos-
sibility that politicians whose political careers and fortunes are built outside
the legislative party system occupy the chief executive office, and they may
use their influence and authority toward ends at odds with legislative voting
discipline (Linz 1994). Most presidents have substantial constitutional
powers over lawmaking, many have extensive controls over the legislative
agenda, and many also control substantial fiscal resources that legislators
covet (Shugart and Carey 1992; Carey and Shugart 1998; Baldez and Carey
134
1999; Figueiredo and Limongi 2000; Aleman and Tsebelis 2005). Presi-
dents also command the attention of national media, and Calvo (2007)
demonstrates that strong public approval enhances presidents’ ability to
sway legislators toward their proposals.
Given their considerable resources, where presidents’ demands contra-
dict those of legislative party leaders, their influence should undermine
voting unity in systems with popularly elected presidents relative to pure
parliamentary systems. To the extent that presidents might use their
resources to pull votes from any and all legislators, the effect should be
present across all parties.
H3: Party unity should be lower in systems with popularly elected pres-
idents than in pure parliamentary system.
Next consider the difference between regimes with presidents and those
without with respect to how a party’s status, in government or in opposi-
tion, ought to affect party unity. The resources of the executive branch
reinforce the influence of legislative party leaders in the absence of a pres-
ident, but they can undermine this influence if vested in an independent
president. Consider first the no-president scenario. In parliamentary sys-
tems, the party leadership is the principal most influential over any given
legislator, and in the case of government parties, the legislative party lead-
ers and the executive are one and the same. Where legislative leadership
and executive leadership are fused, parties in government have more
resources to impose discipline than do those in opposition (Laver and
Shepsle 1996). This suggests:
H3a: Party unity should be higher in governing parties than opposition
parties under parliamentarism.
Under presidentialism, and in hybrid regimes with elected presidents,
the situation is more complex. The general story from Hypothesis 3, that
presidents compete with party leaders for legislator loyalties, applies, but
presidential attention and influence fall more heavily on the presidents’
copartisans and coalition partners than on legislators from outside the
presidents’ coalition. During interviews in various presidential countries,
I asked, ‘‘On what are voting coalitions based – common ideology, party,
electoral interests, control of the legislative agenda, support for the exec-
utive, etc.?’’ Party was the most commonly cited foundation for legislative
voting, consistent with the premise that legislators are beholden to legis-
lative party leaders. Next most frequent, however, were responses that
135
pointed toward the executive. Oscar Hernández (interview) suggested that

legislative party strength in Costa Rica hinges on a party’s relationship to
the presidency:
When a party wins, that party group generally forms a stronger connection to the
executive. The strongest relationship is legislative group-to-executive – president of
the Republic, ministers and all the apparatus of public administration. The losing
party group does not maintain much of a strong connection with its party organi-
zation. Parties in this country are not strong ideological structures, such as would
elicit discipline from each deputy. Parties at the national level have been converted
into electoral platforms more than the classical concept of an ideological bloc.
The interviews illustrated the mechanisms by which presidential

resources – budgetary and regulatory resources and often the ability to
influence the legislative agenda directly – are employed to pressure
copartisans and to build legislative majorities. For example, according
to Nicaraguan deputy Luis Urbina (interview), of President Arnaldo
Aleman’s Liberal Party,
The executive normally works better when it has an assembly majority. The
majority party tries to support projects from the executive, of which it is part. . . .
So when the executive wants to submit a law, it calls on the majority party, explains
the benefits of approving the law, and generally we vote in line with the directives we
are given. This doesn’t mean we are obliged to, but normally that’s what happens.
Urbina’s account relies on an inherent compatibility between the exec-

utive’s interests and those of his legislative copartisans, or at least on some
inherent loyalty to the executive. More frequently, legislators pointed to
concrete resources by which executives elicit support. Critics of Urbina’s
Liberal Party in Nicaragua, for example, invariably pointed to patronage as
the source of support for the executive (Baltodano and Hurtado interviews).
As Jorge Samper (interview) of the Sandinista Renewal Movement put it,
The Liberal group has been very obedient, through presidential discipline more
than party discipline. They take almost no decisions autonomously from the pres-
ident, and when they have done so, they have had to backtrack when it produces
a presidential veto. One or another deputy has voted against the president’s wishes,
and then along comes some bit of patronage that makes him change his vote, and we
vote again the way the president orders. [Interviewer: What are examples of patron-
age?] Public jobs, for deputies and relatives. The deputy might be made ambassador
to some country, and maybe they send his family or relatives. . . . The rest of the
deputies, that are not from the Sandinista or Liberal groups, many of them have
formed alliances with the government . . . [but] these aren’t real political alliances,
but rather alliances based on patronage.
136
Samper’s account of presidential influence is consistent with Nicaragua’s

reputation, and that of the Aleman administration in particular, for cor-
ruption. Accounts of presidential influence elsewhere do not always hinge
on exchanges that reek so much of impropriety, but they share a focus on
the resources executives control that appeal to legislators’ ambitions.
Ricardo Combellas (interview) cites President Chavez’s control over party
nominations to all electoral posts as the main source of his influence within
the Movimiento Quinta República (Chavez’s party in the early years of the
2000 decade) in Venezuela. Carlos Vargas Pagán (interview) cites the Costa
Rican president’s ability to expedite, or hold up, disbursement of funds
budgeted for projects in deputies’ districts as a source of influence.
There is a consensus in the interview responses that presidents control
resources that can be employed to influence legislative votes. The inter-
views raise some questions, however, about exactly how presidential in-
fluence should manifest itself in party unity. First, the accounts of
presidential influence offered in the interviews rely on executive-legislative
exchanges that – even if not overtly corrupt – might attract criticism if
exposed to public scrutiny. This suggests that presidential influence might
be mitigated by visible voting. The obstacle to testing this proposition
empirically is that where votes are not recorded, party unity cannot be
measured, so it remains conjecture. Second is the matter of whether
presidential influence ought to raise or diminish party voting unity. The
Hernández and Urbina interviews suggest that party unity is boosted when
a party holds the presidency relative to some baseline level. This outcome
may be true in the cases of the parties and presidents Hernández and
Urbina had in mind, but whether it is the effect of alliance with presidents
more generally is not obvious. Presidential influence ought to depend on
whether the legislative party leadership consistently agrees with the pres-
ident’s wishes or, if it does not, on the relative influence of each of these
actors over the legislative rank and file.6
More specifically, when the two principals of governing-party legislators
(party leaders and the president) concur on a given measure before the
6
For example, Weldon (1997) demonstrates that the source of unity in the archetypal case of
airtight party discipline under presidentialism – Mexico’s Partido Revolucionario Institu-
cional during its long hegemony – was the elaborate institutional structure that afforded
presidents not only their constitutional authorities but also control over all the partisan and
procedural resources that, in other political systems, normally fall in the domain of legis-
lative party leaders. The secret of PRI unity throughout much of the twentieth century was
that legislators had only one meaningful principal.
137
Table 6.2. Legislators’ Principals under Presidentialism, Parliamentarism,

Government, and Opposition
Opposition Parties Government Parties

Parliamentarism Legislative party leadership Legislative party leadership fused
with executive authority and
resources
H3a: Reinforces party unity relative
to opposition parties
Presidentialism 1. Legislative party leadership 1. Legislative party leadership
2. Some independent pressure 2. Heightened independent
from president influence from president
H3: Undermines party unity H3b: Undermines party unity
relative to all parties in pure relative to governing parties under
parliamentary systems parliamentarism
assembly, the effect should be similar to that under parliamentarism –

providing a boost to unity owing to the additional resources with which
the president can pressure legislators. On the other hand, when the pres-
ident and legislative party leadership disagree and pull in opposite direc-
tions, party unity should suffer in governing parties. Whether the net effect
of competing principals is to diminish voting unity among governing par-
ties under presidentialism depends on how frequently the principals pull in
opposite directions and on their relative influence over legislators. To the
extent the principals compete at all, however, voting unity in government
parties should suffer under presidential systems relative to parliamentary
systems.
All of this suggests the following effect of alliance with the president on
party voting unity:
H3b: Party unity in governing parties should be lower under presiden-
tialism than under parliamentarism.
The logic of H3, H3a, and H3b is summarized in Table 6.2.
6.4.3. Further Hypotheses on Party Unity

Beyond those generated specifically by the logic of competing principals,
a few more hypotheses merit consideration – one that focuses on the
138
parliamentary-presidential distinction, and two that focus on the venera-

bility of regimes and parties and its potential effect on unity.
Among the most prominent propositions regarding the effects of formal
institutions on party unity is that the authority of the executive in parlia-
mentary systems to offer legislative proposals as matters of confidence
accounts for more unified parties in parliamentary than in presidential
systems. The intuition is that the confidence provision raises the stakes
for all parties because rejecting such a measure triggers the collapse of
the government and perhaps early elections. If a party splits, and loses as
a result, on a vote subject to a confidence provision, the costs are greater
than just forgoing the new policy for the status quo, or vice versa (Bagehot
1867; Cox 1987; Diermeier and Feddersen 1998; Huber 1996). This
implies the following hypothesis.
H4: Party unity should be higher in systems with confidence vote pro-
visions than those without.
The confidence vote story is compelling but not without proviso. First,
even where confidence vote provisions exist, they are not formally sum-
moned on most votes, so technically there is room for party voting disunity
that does not threaten government survival. More important, note that the
confidence vote is not restricted to pure parliamentary systems but is a char-
acteristic of hybrid regimes as well (Valenzuela 1994; Shugart and Carey
1992). The best-known case combining a confidence vote provision for the
cabinet with a more-than-ceremonial elected presidency is the French Fifth
Republic (1958–present), but such hybrid arrangements are common among
newer regimes (Frye 1997). In short, the distinction between regimes with
and without confidence vote provisions does not map perfectly onto the
distinction between those with and without elected presidents. In principle,
this ought to allow leverage to distinguish between H3 and H4, to determine
whether differences in voting unity between pure presidential and pure par-
liamentary regimes are the product of competing principals, confidence
votes, or some combination of the two. In practice, scarcity of data from
hybrid regimes hampers my ability to test for that distinction, for now. I
return to this matter in the discussion of the statistical model and results.
Next, consider the age of the regime. Scholarship on comparative party
systems posits that parties in new democracies are weaker than those in better-
established systems (Mainwaring and Scully 1995; Coppedge 1998; Carroll,
Cox, and Pachon 2006). It follows that legislators’ expectations about which
parties will thrive are less solid and that their commitment to any particular
139
party’s collective reputation should be lower in new rather than established

political systems. There should be diminishing returns to the effect of time on
expectations, such that the difference between a regime that is one year old
and one that is eleven years old should be greater than that between one that is
eleven and one that is twenty-one, and so on. This suggests:
H5: Party unity should increase with the age (logged) of the political
regime.
Although new political regimes frequently give birth to new parties and
party systems, such that the age of parties is strongly correlated with the age
of the regimes in which they operate (p ¼ .66), the two remain distinct. New
parties are occasionally born and take root in established regimes, such as
the U.S. Republicans in the 1850s or Israel’s Shas in the 1980s. By the same
token, established parties frequently survive through authoritarian interludes
and thrive after the reestablishment of a democratic regime, such as Argen-
tina’s Radicals (UCR) and Peronists (PJ) in the 1980s or the Christian Dem-
ocrats in Chile and in the Czech Republic in the 1990s. By the same logic as
in H5, legislators’ expectations about the future value of a party’s label ought
to be strengthened the better established that label is and the more durable it
has proved to be over time (Roberts and Wibbels 1999; Stokes 2001).
H6: Party unity should increase with the age (logged) of the party.
Finally, formal models of party competition represent the inherent level
of agreement among copartisans, described here as cohesiveness, as prox-
imity in an N-dimensional policy space (Shepsle 1991). If legislative party
groups comprise ideological neighbors, then for a policy space of a given
size and dimensionality, greater party fragmentation would imply more
cohesive party groups (Cox 1990). Given that cohesiveness is a key source
of voting unity, it follows that party voting unity may be higher in more
fragmented party systems.
H7: Party unity should increase with fragmentation of the legislative
party system.
Among the factors posited here to affect legislative voting unity, some
(e.g., existence of a presidency or confidence vote, the electoral system, fed-
eralism, regime age, or the fragmentation of the party system) are fixed across
all parties in any given assembly, whereas others (e.g., government or oppo-
sition status, party age) vary across parties within the same assembly. Table
6.3 presents descriptive statistics for the system-level variables for the
140
Picturing Party Unity across Systems
Table 6.3. System-Level Variables for Lower Legislative Chambers, by Country
Country Intraparty Federal President Confidence Regime Effective

Competition Vote Age (log) Number
of Parties
Argentina No Yes Yes No 0.69–2.56 2.21–3.59
Australia No Yes No Yes 4.56 2.61
Brazil Yes Yes Yes No 1.61–3.00 4.48–9.30
Canada No Yes No Yes 4.86 2.26
Chile No No Yes No 1.95–2.20 4.85–5.12
Czech No No No Yes 0.69–1.61 3.80–5.90
Republic
Ecuador No No Yes No 2.94 5.05
France IV No No No Yes 0–2.40 4.30–6.76
Republic
Guatemala No No Yes No 2.30–2.71 2.16–3.46
Israel No No No Yes 3.91 8.50
Mexico No Yes Yes No 4.28 2.86
New Zealand No No No Yes 5.02–5.04 1.81–2.87
Nicaragua No No Yes No 2.77 2.86
Peru Yes No Yes No 1.79–2.08 2.66–4.01
Philippines Yes No Yes No 3.71 2.15
Poland Yes No Yes Yes 2.20 3.17
Russia No Yes Yes No 1.39 5.66
United States Yes Yes Yes No 5.31–5.34 1.90–2.00
Uruguay Yes No Yes No 0.69–1.95 2.87–3.34
assemblies included in the quantitative analysis in this chapter. In cases of

assemblies where data were available for more than one period, Table 6.3
shows mean values across periods for the effective number of parties (Laakso
and Taagepera 1979) and for regime age (log) during the first year in each
legislature.
6.5. Picturing Party Unity across Systems

The indices developed in Chapter 5 can illustrate cross-national patterns of
voting unity. In the literature on comparative legislatures, the most prev-
alent explanations for levels of unity refer to electoral rules (Wallack,
Gaviria, Panizza, and Stein 2003; Shugart, Valdini, and Suominen 2005)
and to the confidence vote procedure (Bagehot 1867; Cox 1987; Huber
1996; Diermeier and Feddersen 1998; Gerring, Thacker, and Moreno
141
Figure 6.3 Boxplot of WRICEi indices by confidence vote and intraparty

competition.
2005).7 Figure 6.3 presents WRICEi indices for the parties in each country
for which I have data according to whether the constitution includes a con-
fidence vote provision and whether assembly elections provide for compe-
tition among candidates from the same party. In the bottom left panel are
systems with the confidence vote and without intraparty competition. By
and large, voting unity as measured by WRICEi is high, with the average
more than .90. Parties in France’s Fourth Republic are widely regarded to
have been chronically factionalized, but even its mean WRICEi is .85.
Canada and Israel each have a derelict outlier, but in each case these are
two-member parties in which a 1–1 split vote would drive the RICEij score
to zero. Overwhelmingly, the legislators in these parliamentary systems
voted together with their copartisans.
The bottom right panel shows the one case of a confidence vote system
with intraparty competition, Poland. Note that, in addition to the confidence
7
Federalism is frequently mentioned as a source of party disunity at the national level, but in
the most careful analysis of the issue to date, Desposato (2004b) finds evidence for only
a remarkably small drag on voting unity in Brazil, where low party unity had often been
attributed to a federalism effect (Weyland 1996; Mainwaring 1999). Jones and Hwang
(2005) similarly are unable to detect an effect of provincial forces on voting in the Argentine
Congress, although they confront substantial challenges in identifying what effect alliance
with governors ought to have on allied deputies.
142
vote provision, Poland also has a popularly elected presidency. It is the

one hybrid regime from which I was able to collect recorded voting data.8
WRICEi is extraordinarily low. Poland’s open-list proportional represen-
tation may contribute to individualism among members of the Sejm. The
Polish presidency may also contribute to disunity among some parties. It is
worth noting that, of eight parties in the Sejm, President Kwasniewski’s
Social Democrats (SLD) had the lowest WRICEi index, consistent with
the competing principals logic outlined in H3b. We should be cautious
about drawing inferences based on this case, however. The Polish vote data
are from a twenty-month period following the adoption of a new constitu-
tion and the inauguration of a new government facing an opposition presi-
dent. The rules of the game, and the party system itself, were relatively
young, and voting in subsequent periods may show increased unity. Never-
theless, the Polish data at hand are consistent with the propositions that
intraparty competition and the presence of an elected president, as well as
alliance with the president, generate drags on party unity.
The top left panel of Figure 6.3 shows the nonconfidence vote (i.e., pure
presidential) systems without intraparty competition. Nicaragua and Russia
are very low, of course, but WRICEi indices there must be eyed warily in light
of their absolute-majority threshold voting rules. Elsewhere, levels of
WRICEi are higher – a bit lower than under confidence vote systems without
intraparty competition but generally in the .8 to .9 neighborhood. Finally, the
top right panel shows systems without confidence votes and with intraparty
competition, and the indices suggest more modest levels of voting unity over-
all, averaging in the .7 to .8 neighborhood, and with considerable spreads.
Figure 6.4 presents the same set of boxplots for the RLOSERi index, and
here the pattern is similar, although Poland is less extreme. Among the pure
parliamentary cases without intraparty competition, parties almost never
lose votes that they could, but for party cross-voting, have won. In Poland,
the median party lost about 2 percent of all votes because of such divisions.
(It should be noted that this party, the Peasant Party [PSL], was on the
winning side of 92 percent of all votes, so its losses due to disunity
accounted for a quarter of all its losses.) At any rate, caution is again in
8
Attentive readers might note that both the Peruvian (Art. 134) and Russian (Art. 111)
constitutions provide for removal of the cabinet by vote of a parliamentary majority. How-
ever, both constitutions also allow the president to dissolve the legislature in this instance,
raising the costs to legislators enormously of wielding the no-confidence vote over presi-
dential resistance. Given these provisions, I do not code Peru or Russia as no-confidence
vote systems.
143
Figure 6.4 Boxplot of RLOSERi indices by confidence vote and intraparty

competition.
order in drawing inferences about this particular combination of institu-

tional variables from the Polish data alone. The top left panel shows pure
separation-of-powers systems with no intraparty competition, again show-
ing a larger spread of values and slightly greater disunity overall than in the
analogous pure parliamentary cases. Finally, the top right shows the pure
separation-of-powers systems with intraparty competition and, as expected,
exhibits the greatest incidence of lost votes because of disunity. The United
States is the outlier, with a median value of around 9 percent of all votes lost
because of disunity, but values in the 2 to 5 percent range are not unusual in
Peru and Uruguay, and indices run still higher in Brazil.
Analogous boxplot graphs for WUNITYi and ULOSERi are much less
striking, showing larger spreads across the board but no clear patterns.
The next two figures contrast WRICEi with WUNITYi (Figure 6.5) and
RLOSERi with ULOSERi (Figure 6.6), pooling observations from coun-
tries according to whether they have directly elected presidents and intra-
party electoral competition.
In Figure 6.5, the darker boxes represent WRICEi, and the lighter
WUNITYi. There is a clear overall pattern by which WRICEi indices
trend higher in parliamentary systems than in those with elected presidents,
and among presidential systems, lower with intraparty competition than
144
Figure 6.5 Boxplot of WRICEi and of WUNITYi indices by intraparty

competition and presidentialism, pooled across countries.
Figure 6.6 Boxplot of RLOSERi and of ULOSERi indices by intraparty

competition and presidentialism, pooled across countries.
145
without. (There are no recorded vote data from parliamentary systems with
intraparty competition.) There is no analogous pattern among WUNITYi
indices, which exhibit wider spreads and no apparent responsiveness to
regime type or electoral rules. Figure 6.6, with the darker boxes represent-
ing RLOSERi and the lighter ULOSERi, shows a similar contrast, with
larger increasing RLOSERi values from parliamentary to presidential sys-
tems, and from no intraparty competition to intraparty competition within
presidential systems, but no corresponding pattern for ULOSERi. On the
whole, the mobilization-based indices, WUNITYi and ULOSERi, appear
to be more susceptible than the cross-voting-based indices, WRICEi and
RLOSERi, to distortion via nonvoting equilibria, as discussed in Chapter 5,
which limits their usefulness for cross-national comparisons.
On the whole, the graphical patterns of WRICEi and RLOSERi suggest
that party unity is lower in presidential regimes and in the presence of intra-
party electoral competition. The statistical analysis below adds variables, op-
erating at both the level of political system and the level of individual parties,
to gain additional purchase on sources of party voting unity, and disunity.
6.6. Models
6.6.1. Challenges Presented by the Data

The structure of the data presents some challenges in testing the hypotheses
developed here. Each observation represents the characteristics of a party
group in a given assembly. Multilevel modeling is appropriate for the multi-
variate analysis because each party group is clustered within a higher-level
unit, a legislative assembly. Factors like party age or government-opposition
status that vary across parties within each assembly are posited to affect voting
unity, but the hypotheses also refer to system-level factors, like presidential-
ism or electoral rules, that are constant across parties within a given assembly
but vary across assemblies (Singer 1998; Steenbergen and Jones 2002).
Another data issue is covariance between two of the system-level varia-
bles. The data include only one hybrid constitution, combining a popularly
elected and powerful president with a confidence vote provision for the
cabinet – that of Poland. Otherwise, presidentialism and the existence of
a confidence vote procedure are perfect complements of each other, mak-
ing it difficult to distinguish their effects statistically. Ideally, vote data
from more hybrid systems will become available, so these factors can vary
independently to a greater degree. In the meantime, multicollinearity
146
Models
means it is not feasible to include both a Confidence Vote variable and

a Presidentialism variable in the same regression estimating voting unity.
The results presented here include the Presidentialism variable for two
reasons. First, testing H3b requires including a cross-level interactive vari-
able, Government Party * Presidentialism, and there are compelling sta-
tistical reasons to include the component variables when using an
interactive variable (Brambor, Clark, and Golder 2006).9 Second, in some
specifications, Presidentialism shows a statistically significant impact on
voting unity, whereas the Confidence Vote variable never does. I report
this latter (non)result in the text but show the statistical results of the
models including the Presidentialism variable.
6.6.2. The Multilevel Model

The basic statistical model is to estimate voting unity as a function of both
party-level and system-level explanatory variables, as follows:
UNITY ij ¼ p0 j þ p1 P1 þ ::: þ p4 P4 þ eij
p0j ¼ b00 þ b01 S1 þ ::: þ b5 S5 þ r0j
P1–P4 and S1–S5 refer to party-level and system-level variables. The

party-level coefficients (p1–p4) are estimated as fixed across countries. The
party-level intercept (p0j) is a function of both party-level and system-level
effects. To estimate the cross-level interaction, government party status by
presidentialism (per H3b), a party-level coefficient is modeled as a function
of system-level variables.
The dependent variables are the two cross-voting-base measures of vot-
ing unity developed in Chapter 5, WRICEi and RLOSERi, as well as the
parties’ overall averages (%WONi) on recorded votes.10
Note that all three voting unity indices under examination are con-
strained to values between zero and one, with WRICEi fairly normally
9
I thank an anonymous Cambridge University Press reviewer for pointing me toward the
Brambor, Clark, and Golder (2006) article, which illustrates the importance of including all
component variables along with any interactive term. An earlier version of this study (Carey
2007), not fully appreciating this imperative, presents models including Confidence Vote,
rather than Presidential, alongside the Government Party * Presidential interactive term.
10
Analyses on the mobilization-based indices, WUNITYi and ULOSERi, are not included
in the analyses presented in this chapter because the sensitivity of those indices to non-
voting makes them less reliable measures of unity, at least until effective means of identi-
fying and accounting for nonvoting equilibria can be established.
147
distributed around a mean of .79; %WONi fairly uniformly distributed

with mean .59; and RLOSERi, clustered around zero, with mean ¼ .014
and standard deviation ¼ .032. The assumptions of the linear model are
under the most serious strain for the analyses of RLOSERi, given its clus-
tering to the left of the scale. To verify the results from the linear model, I
replicated the analyses using multilevel ordered logit, grouping ranges of
values on the dependent variables into bins where necessary. All significant
results from the linear models are confirmed in the ordered logits, and in
some instances, the logit coefficients are significant where the linear models’
estimates fall just short. I report on the linear multilevel models for sim-
plicity of exposition and interpretation of coefficients, noting on occasion
where the ordered logits confirm hypotheses that are ambiguous in the
linear models.
System-level explanatory variables include the following:
Intraparty Competition is coded 1 if the electoral system requires that
candidates for the assembly compete against their own copartisans
for preference votes; 0 otherwise.
Presidential is coded 1 if the country has a popularly elected presidency
endowed with substantial constitutional powers; 0 otherwise.
Federal is coded 1 if the country has a federal constitution and subna-
tional units are meaningful arenas of political competition and the
distribution of political resources; 0 otherwise.
Regime Age (log) is the log of the number of years since the founding of
the current democratic regime.
Effective Number of Parties is the Laakso and Taagepera (1979) index of
party system fragmentation for a given chamber.11
Party-level variables are as follows:
Party Age (log) is the log of the number of years since a given party’s
founding.
11
Note that both the Effective Number of Parties and the previous system-level variable,
Regime Age, can vary across assemblies within a given legislature. From Table 5.1, for
example, note that the data include votes from three different Brazilian assemblies, two
from Chile, but one from Australia. Potentially, then, there are three levels of data – party
within an assembly, the specific iteration of the assembly, and the constitutional system in
which the iterations of that assembly exist. In the interest of simplicity, however, I combine
these last two levels, assigning mean values for the Effective Number of Parties and Regime
Age across observations from a given legislature. In practice, there is little variance on
either variable across assemblies within each system for the data included.
148
Models
Seat Share is the proportion of seats held by the party in a given assembly.
Government Party is coded 1 if the party holds at least one cabinet port-
folio in the current cabinet; 0 otherwise.
Government Party * Presidential is the interaction of Government Party
with Presidential.
The logic of the independent variables and expectations about their
effects are mostly straightforward from the hypotheses section, but a few
comments are in order. The default status implied by the models is for an
opposition party in a parliamentary system. The variable Presidential esti-
mates the effect on the dependent variable of being in a presidential regime.
The coefficient on Government Party represents the effect of being in
government, as opposed to out, in parliamentary systems. The coefficient
on Government Party * Presidential picks up the marginal difference be-
tween government parties in systems with directly elected presidents and
those in parliamentary systems. The comparison between government
and opposition parties within presidential systems is also of interest, so
below the list of coefficients from each model the tables show the linear
combination of Government Party + Government Party * Presidential –
Presidential.
Seat Share is included as a control variable, but its logic depends on
the dependent variable. When the dependent variable is %WONi, expec-
tations regarding Seat Share are clear-cut – a greater share of seats should
lead to more wins. When the dependent variable is WRICEi, expectations
are less firm. Parties that comprise larger shares of their chambers may
encompass more-diverse viewpoints and thus be subject to disunity. On the
other hand, increasing seat shares generally provides increasing access to the
legislative resources that party leaders employ to elicit compliance and to
mobilize their rank and file (Hurtado interview). Finally, when the depen-
dent variable is the RLOSERi index of vote losses due to disunity, the effect
of Seat Share should be positive, notwithstanding the fact that bigger par-
ties win more, because a split within a larger party should be more likely to
reverse a vote outcome than the same split in a smaller one.12
12
I also ran the models on vote loss due to disunity controlling for WIN%, on the grounds
that only parties that win votes stand to lose some through breakdowns in unity. That is, if
a party’s winning percentage is zero or close to it, we might reasonably expect that it is
merely in perpetual and futile opposition, rather than that it might have won, say, 3 percent
of those lost votes but for internal splits. This turns out not to be the case, however; the
coefficient on WIN% was never close to significant.
149
6.7. Results
Models 1a–c, in Table 6.4, regress WRICEi on sets of both system-level
and party-level explanatory variables. Model 1a includes the full set of
variables implied by the preceding hypotheses, plus a control for each
party’s share of assembly seats. It is, perhaps, most straightforward to note
at the outset the hypotheses for which there is no support in any model.
Contrary to H2, there is no evidence that federalism undermines party
voting unity, as measured either by WRICE or by one of the vote-outcome
(%WON or RLOSER) indices. Nor is there evidence that the existence
of a confidence vote provision (H4), regime age (H5), or party system
fragmentation (H7) matters under any model specifications. With regard
to the confidence vote, statistical results are not shown in Table 6.4 because
Confidence Vote cannot be included in the same model with Presidential,
but alternative specifications dropping Presidential (and Government
Party * Presidential) show no measurable effect of Confidence Vote on
any of the dependent variables.
6.7.1. Party Age Fosters Unity

In Model 1a, the coefficients on all system-level variables fall short of
conventional significance levels (although that on Presidential just barely,
with p < .12). The only party-level variable with any leverage is Party Age
(log). The effect of party longevity is modest but consistent. An increase in
one natural log unit (say, from a 1-year-old party to a 3-year-old party, or
from 3 to 8 years, 8 to 20, 20 to 55, or 55 to 150) is expected to increase
WRICEi by .02. Moving from the youngest parties in the data (newborns
populate many assemblies) to the oldest (e.g., the major parties in Uruguay
or the United States) is expected to push WRICEi up by .10, or half
a standard deviation, other things being equal. To the extent that legislative
voting unity reflects a party’s brand name, such reputations gradually grow
clearer and more informative over time. None of the other party-level
variables provides any traction in explaining WRICE. Neither a party’s
share of seats nor its status in government or out has any measurable effect.
6.7.2. Intraparty Competition and Presidents Disrupt Unity

The effects of two system-level variables of interest, however, warrant
further exploration. With usable data from only nineteen assemblies, there
are substantial degrees of freedom constraints at the system level.
150
Table 6.4. Multilevel Analysis of Legislative Voting Unity within Parties
Dependent Variables
WRICEi %WONi RLOSERi
Independent Variables 1a 1b 1c 2 3a 3b 3c
System level
Intraparty Competition .04 (.10) .05 (.10) .16** (.07) .09 (.07) .017* (.010) .018* (.009) .021** (.010)
Presidential .17 (.11) .19* (.10) .06 (.07) .14* (.08) .004 (.010) .002 (.010) .003 (.011)
Federal .02 (.09) .12** (.06) .004 (.009)
Regime Age (log) .02 (.04) .02 (.03) .003 (.004)
Effective Number Parties .01 (.03) .01 (.02) 000 (.003)
Party level
Party Age (log) .02*** (.01) .02** (.01) .02** (.01)
Seat Share .05 (.08) .05 (.08) .01 (.07) .38*** (.11) .035*** (.008) .036*** (.008) .036*** (.008)
Government Party .05 (.04) .05 (.04) .05 (.03) .25*** (.05) .001 (.003) .001 (.003) .001 (.003)
GovPty * Presidential .01 (.05) .01 (.05) .00 (.04) .18*** (.06) .010** (.004) .010** (.004) .012*** (.004)
Constant .77*** (.21) .89*** (.08) .90*** (.05) .51*** (.15) .010 (.020) .001 (.007) .001 (.007)
H0: Government Party + GovPty * .13 (.11) .15 (.11) .01 (.08) .07 (.09) .007 (.011) .009 (.010) .015 (.011)
Presidential – Presidential ¼ 0
Proportion of variance explained
System level .31 .28 .60 .39 .47 .45 .52
Party level .05 .05 .06 .19 .22 .22 .24
N (parties) 206 206 184 309 254 254 229
N (assemblies) 19 19 16 19 19 19 16
Note: Models 1c and 3c exclude observations from Russia, Nicaragua, and Guatemala, where nonvotes are counted as nay votes. * Significant at > .10
151
level. ** Significant at > .05 level. *** Significant at > .01 level. All models include only parties with two or more legislators.
Therefore, Models 1b and 1c (as well as 3b and 3c) drop those system-level
variables for which there is no hint of a measurable effect on party unity,
retaining those for which the graphical representation of data, as well as
bivariate analyses, suggests some impact. Model 1b replicates 1a, but with-
out the Federal, Regime Age, and Effective Number of Parties variables.
The estimated impact of presidentialism on voting unity increases slightly,
and the standard error shrinks, such that the coefficient is now significant at
.06. The shift from a parliamentary regime to one with an elected president
drops WRICEi by .18 – nearly a full standard deviation. This suggests
confirmation of H3, yet sorting out the relative impact of presidentialism
and intraparty electoral competition remains thorny.
Note that among these data, all cases of intraparty competition are in
regimes with elected presidents. Recall further, from Chapter 5 (Section
5.3.2 and Appendix 5), that applying the WRICEi index to assemblies where
all nonvotes are counted as nays is particularly awkward and may yield
unduly low values. All three such assemblies in the data, those of Nicaragua,
Guatemala, and Russia, are in presidential systems without intraparty com-
petition (see Figure 6.3). Model 1c drops these observations, with a result-
ing shift in the relative scope, and the statistical significance, of the
Presidential and Intraparty Competition variables. With the nonvote ¼
nay-vote observations excluded, Intraparty Competition is associated with
a .16 drop in WRICEi, significant at .02. The estimated effect of presiden-
tialism remains negative but falls short of significance (analogous to Intra-
party Competition when the nonvote ¼ nay-vote assemblies are included).
In short, presidentialism and intraparty electoral competition both appear
to diminish WRICEi, but although the cross-national breadth of these data
is unprecedented, statistical leverage at the system level is limited, so un-
certainty as to the relative impact of these factors remains.
6.7.3. Overall Averages: Governing Parties Win, but Presidents Get

in the Way
Model 2, estimating %WON, begins to shed light on the differences between
governing parties in parliamentary and presidential systems. Note that all
nonconsensual votes pit some winners against some losers, and the hypoth-
eses derived previously do not suggest that the proclivity of parties to be on
the winning side should differ according to system-level factors. Neverthe-
less, to generate as clear a picture as possible of what characteristics of parties
contribute to winning and losing votes, Model 2 regresses %WONi across all
152
Results
recorded votes on both system-level and party-level explanatory variables.13

The system-level coefficients indicate that legislative bandwagons are more
prevalent in presidential than parliamentary systems and less pronounced in
federal than unitary systems, suggesting a greater proclivity among parties in
parliamentary and federal regimes to take a stand – even a losing one.
Moving to party-level factors, the coefficient on the Seat Share control
variable is positive and significant. For every additional percentage of cham-
ber seats a party holds, its expected winning rate rises by nearly four-tenths
of a percent. It is, of course, not surprising that bigger parties tend to win
more. Turning to the variables of greater interest, the coefficient on Gov-
ernment Party shows that, in parliamentary systems, government parties
win at a much greater rate than do opposition parties – 25 percent more,
over and above the effects of Seat Share. The coefficient on Government
Party * President shows that this advantage is largely wiped out for govern-
ing parties under presidentialism – and the difference across systems in the
effect of governing party status is strongly significant. Just as striking is the
linear combination of regime-type governing status, and their interaction,
shown in the next section. Accounting for seat share, the legislative averages
of governing parties in presidential systems are statistically indistinguish-
able from – and in the event slightly lower than – those of opposition parties.
6.7.4. Disunity Losses: Intraparty Competition and Presidents (Again)

Why are government parties in presidential systems not more effective at
winning votes? Model 3 indicates that a substantial share of losses is attribut-
able to breakdowns in unity. Beginning, as previously, with the system-level
variables, neither federalism, presidentialism, regime age, party system frag-
mentation, nor the existence of a confidence vote provision (not shown) has
any measurable effect on RLOSERi’s failing to confirm Hypotheses 2, 3, 4,
or 5. The estimated coefficient on Intraparty Competition is positive and
significant at .07 or better across all three models. RLOSERi jumps by around
2 percent (standard deviation is 3 percent) with Intraparty Competition, in
13
In analyses of %WON and RLOSER, the Party Age variable is never significant. Its in-
clusion or exclusion does not affect the direction or significance of any other variables, but
because I was unable to determine Party Age for almost a third of all parties included in the
data, including the variable in any analysis ‘‘costs’’ a lot of observations. Therefore, Party
Age is dropped from Models 2 and 3. The system-level variables are included as controls in
Model 2, although they are not of substantive interest. The effects of party-level variables
are the same whether the system-level variables are included or not.
153
further support of Hypothesis 1. When copartisans compete for electoral

support, their inclination to distinguish themselves from each other evidently
outweighs party loyalty in some instances that are pivotal to vote outcomes.
Moving to party-level effects, the Seat Share variable is a strong and
significant contributor to RLOSERi rates, confirming that splits in large
parties are more consequential to vote outcomes than analogous splits in
small parties. There is also further support in Model 3 for Hypothesis 3b.
First, note that governing parties in parliamentary systems lose votes be-
cause of breakdowns in unity at effectively the same rate as opposition
parties, which, in practice, is to say only very rarely. The estimated effect
of being in government in systems with presidents, by contrast, is substan-
tial. Governing parties in systems with presidents lose because of disunity
a full percent more than do governing parties in parliamentary systems.
They even lose votes they could have won, but for cross-voting, at higher
rates than opposition parties in presidential systems.14
An additional 1 percent more losses – whether owing to intraparty
competition or governing status in a presidential regime, may appear to
be only a moderate disadvantage, but consider that the mean RLOSER rate
across all parties is 1.4 percent. Governing parties under presidentialism
lose far more than their fair share of votes because of disunity. Further-
more, governing parties, even under presidentialism, win 82 percent of all
votes (as against 84 percent for governing parties in parliamentary systems),
so membership in government can be expected to boost a party’s overall
rate of floor losses by around 5 percent through its effect on RLOSER
alone, accounting for half the overall difference in the governing parties’
averages between presidential and parliamentary systems.
6.7.5. Illustrative Cases

Returning to the ‘‘snapshots’’ of party systems in specific assemblies illus-
trates the overall results. Figure 6.7 shows unity indices based on votes
from the Canadian House of Commons in 1994–97. In this federal system
with no intraparty competition, a confidence vote provision, and no elected
president, WRICEi is uniformly near perfect. WUNITYi is lower but
strongly correlated across parties, suggesting nonvoting equilibria, and
RLOSERi and ULOSERi levels are near zero.
14
The coefficient on the linear combination of regime type and governing status falls short of
significance in the models shown in Table 6.4, but is positive and significant at .01 in the
ordered logit version of Model 3c.
154
Results
Figure 6.7 WRICEi, WUNITYi, RLOSERi, and ULOSERi indices for Canada,
1994–97.
The case of Brazil during the second administration of Fernando

Henrique Cardoso, 1999–2003, is shown in Figure 6.8.15 The president’s
Social Democratic Party (PSDB) is shown in black, second from the left,
while other parties in the government coalition are in white, with the
opposition in various gray hues. Voting unity is lower by all indices among
the governing parties, which also experience markedly higher loss rates
because of to cross-voting and undermobilization.
Those familiar with Brazilian politics will note that among the more
highly unified opposition parties is the Worker’s Party (PT) led by Luiz
Ignacio (Lula) da Silva, which has long been noted for its strong discipline
even within Brazil’s famously fractious and fluid party system. Lula won the
presidential election following Cardoso’s second term, bringing the PT to
15
Indices from Cardoso’s first administration were shown in Figure 5.8, and the picture is
strikingly similar.
155
Figure 6.8 WRICEi, WUNITYi, RLOSERi, and ULOSERi indices for

Brazil, 1999–2003.
power in coalition with a left-leaning bloc of parties. Given their prior

reputations, one might expect these parties to be more unified than their
predecessors in government, and Figure 6.9 shows this to be correct for
votes during Lula’s first two years, 2003–5. Note, however, that the PT’s
WRICEi during the previous two periods was .98, but it fell to .91 even
during Lula’s honeymoon, and qualitative accounts suggest that divisions
within the PT, between legislators loyal to Lula’s and those who objected
to the president’s centrist governing strategy, drove PT unity levels down
further still as Lula’s presidency wore on (Fleischer 2004).
6.7.6. Summing up the Results on Party Unity

Table 6.5 summarizes the empirical evidence regarding the seven hypoth-
eses on voting. Most, but not all, of the competing principals hypotheses
156
Results
Figure 6.9 WRICEi, WUNITYi, RLOSERi, and ULOSERi indices for

Brazil, 2003–5.
find some support. H1 is supported insofar as intraparty electoral compe-

tition is associated with lower values of WRICEi (the effect is statistically
significant when assemblies where nonvotes ¼ nay votes are excluded), and
higher values of RLOSERi. Where electoral rules provide for intraparty
competition, thus strengthening the influence of personal vote constituen-
cies relative to party leaders, there is more cross-voting, leading to in-
creased vote loss rates. H2 fares less well, with no support from the
statistical analyses for the proposition that federalism affects party unity.
The graphical comparisons of voting unity indices across regimes with and
without elected presidents suggest that presidentialism disrupts party
unity, consistent with H3, and there is some statistical support for this
proposition, although the effect is significant only when assemblies where
nonvoting ¼ nay are included in the analysis.
Both H3a and H3b, regarding the relative impact of being in govern-
ment in systems with and without elected presidents, find some support in
157
Table 6.5. Summary of Hypothesis Tests
Hypothesized Effect WRICEi %WONi RLOSERi

Competing principals hypotheses
H1: Intraparty electoral competition reduces unity (U) U
H2: Federalism reduces unity
H3: Presidentialism reduces party unity (U)
H3a: Parliamentarism strengthens governing parties U
H3b: Governing parties weaker under
presidentialism than parliamentarism U U
Other hypotheses
H4: Confidence vote increases unity
H5: Parties more unified in longer-established regimes
H6: Older parties more unified than newer ones U
H7: Parties more unified in fragmented party systems
the results on vote outcomes. In parliamentary systems, governing parties

win more than their share of votes (%WON). Governing parties in pres-
idential systems, by contrast, win at lower rates, and when they lose, they do
so more frequently due to cross-voting. Presidentialist governing parties, in
fact, appear to be at no legislative advantage even relative to their own
opposition. Controlling for Seat Share, they win at no higher rate and
suffer disunity losses more often.
Table 6.5 summarizes the evidence for each hypothesis against
WRICEi, %WONi, and RLOSERi. Check marks indicate that the hypoth-
esis finds support in the multilevel analysis. The competing principals
hypotheses are listed at the top. All except H2, regarding federalism, find
some support.
The hypotheses based on other, noncompeting principals rationales find
little support. There is no evidence that existence of a confidence vote
affects voting unity, although it is important to acknowledge that the data
here make it difficult to estimate separately the potential effects of the
confidence vote from the effects of independently elected presidents at
the system level. As more legislative voting data become available from
hybrid systems, it will be possible to disentangle these stories. Nor is there
support for the idea that parties are more unified in longer-established
regimes, nor in regimes with greater levels of party system fragmentation.
Older parties exhibit lower levels of cross-voting (WRICEi), making H6
158
Extending the Analysis
the only noncompeting principals hypothesis supported by the multilevel

analysis.
6.8. Extending the Analysis
6.8.1. Governing and Opposition Coalitions

Most party systems do not regularly produce single-party legislative ma-
jorities. Multiparty coalitions are generally necessary to form and sustain
a government under regimes that include the confidence vote provision,
and scholarship on presidential democracies increasingly recognizes the
key role played by multiparty coalitions in organizing executives and
building legislative majorities (Amorim Neto 2002, 2006; Carey 2002;
Powell 2000; Siavelis 2000). The central role of coalitions in legislative
politics suggests that it is worthwhile to take advantage of the flexibility of
the voting unity indices developed here to apply them to governing and
opposition coalitions as wholes, in order to run another analysis comple-
mentary to that on parties themselves.
For the purpose of calculating government and opposition voting unity
indices, I coded all legislators in an assembly from parties that held cabinet
portfolios for more than half of the period from which votes were collected
as being inside the government coalition. All others are lumped into a single
opposition coalition. This means that the opposition ‘‘coalition’’ is fre-
quently not a coalition in any formal sense but rather a residual category
that may well include legislators with widely disparate preferences and
ideological tendencies. There may be organized and internally coherent
opposition blocs in many legislatures – for example, parties that had formed
preelectoral coalitions before the previous election – but because I did not
have information on such blocs across all the assemblies and time periods, I
relied on the crude method of lumping all nongovernment legislators to-
gether. This means the baseline level of unity to be expected within oppo-
sition coalitions is modest.
6.8.2. Adapting the Statistical Model

Having calculated the familiar voting unity indices for government and
opposition coalitions in each assembly, I rely on the familiar multilevel
statistical model, as in the preceding analysis of parties. The independent
159
variables describing political institutions and regime age are the same. I do
not include a measure of party system fragmentation, given that the analysis
includes only a single government and opposition coalition within each
assembly. I do, however, include a coalition-level variable, Multiparty
Coalition, indicating whether a coalition consisted of legislators from more
than one party. The Seat Share variable is the percentage of assembly seats
held by all the parties in a given coalition.
As in the b and c models from Table 6.4, I am parsimonious about
including system-level variables in the models on WRICEi and
RLOSERi to preserve degrees of freedom at that level. I include Intra-
party Competition, which is associated with lower WRICEi indices,
and Presidential, given the prominence of the cross-level interactive
variable, Government Coalition * Presidential (Brambor, Clarke, and
Golder 2006).
6.8.3. Results: Government Bonus under Parliamentarism

but Not Presidentialism
Table 6.6 presents the results of the regressions. As with parties, intraparty
electoral competition diminishes voting unity in coalitions. This may be
a manifestation of the diminution of voting unity fostered by such compe-
tition within the parties that make up broader coalitions, but it may also
reflect a greater level of legislative individualism within such assemblies
generally. The effect is large, negative, and significant at .01 on coalition
WRICEi indices, and positive (as expected) but not significant for RLOS-
ERi. By contrast, presidentialism has no measurable impact on either of the
coalition voting unity measures. All system-level variables, including Fed-
eral and Regime Age, are included as controls in Model 5.
Moving to the coalition-level variables, bigger coalitions experience a bit
more cross-voting (lower WRICEi, per Model 4) than smaller ones, al-
though the estimated effect falls short of statistical significance. Bigger coa-
litions win more frequently (Model 5), as expected, and they lose because of
vote defections more often (Model 6). Multiparty coalitions experience more
cross-voting than single-party groups, win less frequently, and lose because
of defections more, although the latter two effects are not significant.
Of primary interest in the coalition analysis, however, are the conditional
effects of being in or out of government. Governing coalitions in parliamen-
tary systems are more unified than opposition coalitions, with nearly a full
standard deviation boost in WRICEi (.15), a 38-point boost in percentage of
160
Extending the Analysis
Table 6.6. Multilevel Analysis of Legislative Voting Unity within Governing and
Opposition Coalitions
Independent Variables Dependent Variables

WRICEi %WONi RLOSERi
Model 4 Model 5 Model 6
System level
Intraparty Competition .17*** (.07) .10* (.06) .008 (.030)
Presidential .03 (.08) .16* (.09) .002 (.035)
Federal .16*** (.06)
Regime Age (log) .02 (.02)
Party level
Seat Share .22 (.14) .63*** (.17) .010* (.060)
Multiparty Coalition .14** (.06) .09 (.07) .020 (.025)
Government Coalition .15** (.07) .38*** (.08) .051* (.030)
Government Coalition* .08 (.08) .23*** (10) .009 (.035)
Presidential
Constant .91*** (.10) .37** (.15) .012 (.045)
H0: Government .04 (.10) .02 (.12) .044 (.046)
Coalition + GovCoal*
Presidential – Presidential ¼ 0
Proportion of variance explained
System level .53 .60 .04
Party level .31 .55 .21
N (coalitions) 68 68 68
N (assemblies) 14 14 14
Note: Observations from Philippines (no information on governing coalition), Russia
(nonpartisan president), Nicaragua and Guatemala (nonvotes ¼ nays), and Uruguay
(records only from votes in joint sessions on constitutional amendments) excluded.
* Significant at > .10 level. ** Significant at > .05 level. *** Significant at > .01 level.
votes won, and a 5.1 percent reduction in the rate of losses due to cross-
voting – the mean rate of which across all observations is 6.9 percent.
Whereas governing coalitions in parliamentary systems are more unified,
win more votes, and lose less frequently because of disunity than opposition
coalitions, the coefficients on Governing Coalition * Presidential indicate
that this unity boost is diminished in presidential regimes (more cross-
voting, more losses, and more defection losses), although the marginal effect
reaches statistical significance only with regard to winning percentage. The
linear combination of Government Coalition + GovCoal * Presidential –
Presidential (at the bottom of Table 6.6) indicates that governing coalitions
161
in regimes with presidents are statistically indistinguishable from opposition

coalitions in presidential systems on either of the unity indices or on overall
average. Recall, however, that in each assembly, unity indices for opposition
coalitions lump all nongovernment parties together, so the bar for a govern-
ing coalition to surpass is quite low. This makes the result on overall legis-
lative win rates, in particular, all the more striking. As with parties, these
results suggest that presidents are no legislative asset at all to the coalitions
through which they endeavor to govern. In pure parliamentary systems,
government coalitions are more unified and more successful (even control-
ling for Seat Share) in winning votes, but the government unity boost van-
ishes altogether in the presence of presidents.
6.9. Conclusion: Competing Principals Disrupt Voting Unity

The evidence in this chapter supports the competing principals approach
to legislative representation, the basic idea of which is that almost all legis-
lators are subordinate to party leaders within their assembly, and the extent
to which party groups are unified or cohesive depends on whether other
principals, with competing demands, also control resources to pressure
legislators. To the extent that such competing principals elicit responsive-
ness from legislators, they drive wedges into party groups, which we ob-
serve in vote patterns and vote outcomes. This chapter looked for sources
of competition among principals in the constitutional and electoral rules
that govern legislative politics, and in how these institutions interact with
the status of parties inside and outside government.
6.9.1. Electoral Rules Matter

The evidence here supports some arguments, but fails to support others,
about the effects of institutions on legislative party unity that have either
been derived theoretically, or advanced on the basis of evidence from
a smaller number of cases, or both. These results are based on a broader
cross-national dataset than any previous study, which affords greater lever-
age in estimating system-level effects and for disentangling these from
party-level effects. For example, a number of scholars have attributed dis-
unity within parties to intraparty preference voting (Cain, Ferejohn, and
Fiorina 1987; Mainwaring and Pérez-Liñán 1997; Garman, Haggard, and
Willis 2001), whereas others rightly cautioned that, in the absence of ev-
idence from legislative voting itself, inferring levels of party cohesiveness
162
Conclusion
from voting rules alone was premature (Figueiredo and Limongi 2000).
The results here should dispel uncertainty on this count.
6.9.2. No Evidence of a Federalism Effect

Federalism has been identified as weakening national-level parties in case
studies of India and Brazil (Chhibber and Kollman 1998; Mainwaring 1999),
although the most sophisticated studies to date of legislative voting patterns
estimate the effect on party voting unity to be relatively small or undetect-
able (Desposato 2004b; Jones and Hwang 2005). The results here, with
considerably extended empirical reach, find no evidence that federalism
per se affects levels of legislative voting unity. It may be that the blunt
measure of constitutional federalism employed here is insufficiently sensitive
to capture varying levels to which power within national parties is decen-
tralized, or to which there is regional heterogeneity within parties, or both.
6.9.3. Governments Differ in Parliamentary and Presidential Systems

The most important new results are found in the differences between
parliamentary and presidential systems on governing party unity. The dif-
ferences reported here do not rely on the presence or absence of the con-
fidence vote provision, which is at the center of many discussions of party
discipline. Rather, they are based on how the existence and influence of
a popularly elected president can disrupt party voting unity generally, and
an account of how being in government differs in presidential and parlia-
mentary regimes. Take two parties, or two coalitions, of the same size. Put
one in government under parliamentarism, and the additional resources
should boost its legislative effectiveness. By contrast, hand that party or
coalition the presidency, and you should expect no such advantage.
Studies of the presidency in specific countries frequently conclude that
the office is unusually strong, even dominant over the legislature. In the
literature, presidents appear to be an unusually potent breed. The results
here suggest reassessing this verdict, at least with regard to legislative in-
fluence. Parties allied with presidents do not do any better on the floor of
the legislature than others. Presidents may dominate their local political
theaters in various ways but not by directing the actions of unified battal-
ions of legislators.
For all their stature and the resources they command, presidents are
disruptive to party unity because they present a potentially competing
163
source of directives against those of party leaders within the legislature.

Legislative leaders in parties outside government need not contend with
such a formidable competitor in coordinating the actions of their troops.
The incentives for presidents to stake out positions ‘‘above’’ politics and to
carry themselves as suprapartisan actors, even when they have won election
on the basis of party support, buttress this effect. And the resources –
political and material – that presidents command in most systems provide
them ample currency with which to curry legislative favor. By this account,
it is not presidential weakness per se that is the source of party disunity but
presidential power. Power can be understood as a source of party disunity,
however, only if one is attentive to the institutional environment in which
legislative parties operate. The aim of the competing principals theory is to
focus attention on the elements of that environment that shape the strength
of party leaders and the various actors with whom they compete for legis-
lators’ loyalties.
164
7
The Individual-Collective Balance
7.1. Transparency, Party Unity, Votes, and Accountability

This book asks, Who are the political actors in a position to place demands
on lawmakers and, given the mix of pressures, what kind of legislative
accountability can we expect? The focus throughout is on legislators’ votes.
Whatever other important representative and policymaking activities tran-
spire in assemblies, votes remain the core blocks on which legislative
decisions are built. I concentrate on whether votes can be easily monitored
by those outside the legislature – their visibility – and, in those legislatures
where votes are recorded and available for analysis, on patterns of voting
among parties.
These two elements of legislative voting, transparency and party unity,
are key components of two distinct types of legislative accountability: in-
dividual and collective. When the votes of individual legislators are not
visible, it is difficult for those outside the legislature to know whether
a representative has acted in accordance with their preferences. Some mea-
sure of voting unity within groups is necessary for collective accountability
as well, because, if its members do not vote together regularly, a group
cannot be regarded as shaping legislative outcomes.
Transparency and voting unity are matters of degree, not absolutes, and
much of the book is an effort to document and then to explain how much of
each we see across various legislatures. The first task is primarily one of
mapping – of visiting legislatures, personally and virtually, in order to
discover what information about votes is available. Where voting records
can be had, we can expand and improve the map by turning the quantity of
ayes, nays, and nonvotes into statistical descriptors of voting unity. The
explanatory work in the book relies on a combination of ‘‘soaking and
165
poking’’ (Fenno 1978) to determine which political actors favor transpar-

ency and which do not, and developing a model of the forces that play on
legislators, and so affect their proclivities toward party voting unity. I refer
to the actors that apply these pressures as legislators’ principals, and I argue
that party unity is a product of the extent to which these principals pull in
different directions.
The explanations of transparency and party unity are related because
principals that apply competing pressures force a measure of individualism
upon legislators. Deciding how to vote when one’s legislative party pulls
one way and the president another, for example, is an act of self-definition.
If the vote is visible to citizens, then the same act that diminishes party
unity, and so might erode collective accountability, can also provide a build-
ing block for individual accountability.
7.2. Reviewing the Major Points

The ideal of legislative accountability is afflicted by a fundamental
tension between individualism and collectivism.
Individual accountability implies that legislators answer to the specific
demands of citizens in their behavior, including voting. Collective ac-
countability implies that teams of legislators – mainly parties and coali-
tions, in most legislatures – act collectively to promote a policy agenda
and are evaluated by citizens as a group according to their effectiveness in
advancing it. Where constituents – even supporters of the same party or
coalition – put diverse demands on legislators, the demands of individual
accountability can contradict the collective action on which collective
accountability is based.
Academic work on legislative accountability displays a normative pro-
clivity for the collective variety, but there are signs of a push toward
individual accountability in legislatures themselves.
There is a venerable tradition of scholarship on legislative accountability
that extols strong-party government, mainly on the grounds that strong
parties facilitate clear options for voters over policy platforms. Neverthe-
less, political reforms in many Latin American countries have aimed at
boosting individual accountability, even at the expense of strong parties.
Politicians’ survey responses demonstrate a pronounced bent toward
greater individualism. Survey responses may be cheap talk, perhaps
166
Reviewing the Major Points
reflecting only what the politicians think citizens want to hear. Even if so,
though, this would indicate that politicians believe that citizens want
more individualistic and less party-centered representation from their
legislators. What limited evidence is available from public opinion sur-
veys suggests that this is, indeed, what citizens want. At any rate, there
appears to be a disjuncture between what much of the academic literature
on legislative representation prescribes and what politicians and reform-
ers aim to deliver.
Visible votes are an essential component of individual accountability.
They are in scarce supply in many legislatures, but time and technol-
ogy push toward more visible voting.
The mechanics of voting in legislative assemblies can produce a funda-
mental asymmetry in the ability of legislative insiders (most prominently,
legislative party leaders) and outsiders (everyone else, including citizens,
organized interest groups, the media, and academics) to monitor legisla-
tors’ behavior. The historical and institutional contexts differed consider-
ably, but the adoption and expansion of recorded voting in the United
States and in some countries in Latin America have pushed accountability
in similar directions. On the whole, visible voting has been favored by
opposition legislators and has been used to force unpopular measures
advanced by majority parties and coalitions onto the public record. Inter-
views support the idea that recording and publishing legislative votes
facilitate external monitoring, foster fair play in legislative procedures,
discourage legislators from obfuscating their records, allow voters to re-
ward or punish legislators for votes in elections, and may affect legislative
decisions in anticipation of such effects.
The supply of recorded votes is limited in most Latin American legis-
latures, largely because of reluctance about visible voting on the part of
party leaders who strive to maintain a tight grip on the rules of legislative
procedure. Nevertheless, the increasing availability of secure and efficient
electronic voting systems has driven down dramatically the procedural
costs of recording votes and making them visible. Moreover, once some
set of conditions – a presidential initiative, for example, or a successful
minority-led reform – allows for the establishment of visible voting as stan-
dard operating procedure, the practice appears difficult to revoke, perhaps
owing to the widespread belief among politicians that citizens want
more transparency in legislative institutions, even at the cost of strong
party control.
167
Electoral rules matter to what sort of accountability legislators deliver.

The familiar SMD-versus-PR distinction is not what drives the
individualism-versus-collectivism trade-off, but it does affect incen-
tives for candidates to deliver information that facilitates individual
accountability.
The distinction between single-member districts and proportional rep-
resentation is central to so much scholarship on electoral systems that
academic attention frequently gravitates there unreflectively. Yet one finds
highly individualist, and highly collectivist, legislative representation on
both sides of the SMD-PR divide. Electoral rules can encourage, or dis-
courage, individualism, but they do so by shaping the range of principals to
whom legislators respond. The more a centralized national party leader-
ship monopolizes access to the electoral resources legislators value, the
more dominant the party is as its legislators’ common principal, and rep-
resentation is more collective. When electoral resources are more decen-
tralized, legislators of the same party diversify their appeals, responding to
a more heterogeneous group of principals, placing a greater premium on
individualistic representation. The trade-off here has little to do with
single-member districts versus proportional representation and much to
do with the number and nature of principals to which legislators respond.
Although either SMD or PR electoral systems can foster individualistic
(or collective) representation, the number of candidates competing for votes
in a given district affects the electoral advantage to be had by publicizing the
record of one’s electoral opponents. Specifically, the fewer other candidates
competing, the greater the expected gain to any given candidate from crit-
ically exposing an incumbent legislator’s record. The number of candidates
tends to rise with district magnitude. Therefore, this particular transmission
belt for information about incumbent legislators’ voting records – negative
campaigning by other candidates – should be more effective in elections with
low district magnitudes than where magnitudes are higher.
Recording votes, in addition to making individual-level visibility pos-
sible, also makes it possible to measure voting unity across parties,
coalitions, or any group of interest within an assembly.
The indices developed here can be used to generate statistics that describe
the voting unity of any group. Because parties are the universal unit of col-
lective representation in modern democratic legislatures, and because party
unity is widely regarded as a key condition for collective accountability, party
168
The Optimal Mix?
voting unity is of natural interest. I generate statistics to describe parties’

levels of mobilization, cross-voting, their overall success in winning recorded
votes, and their incidence of losses attributable to undermobilization and
cross-voting. These statistics allow for cross-national and cross-temporal
comparisons of party unity, for example, either statistically or graphically.
Other things being equal, the more the institutional context estab-
lishes alternative principals with control over resources legislators
value, the lower is party unity.
The institutional context affects the relative value to legislators of col-
lective party labels versus individualism in voting. For example, longer-
established political parties tend to mobilize their legislators at higher
levels, and levels of cross-voting are lower, suggesting that the communi-
cative value of party labels increases with time. For some institutional
factors, I find no evidence of effects on voting unity. The voting indices
do not support hypotheses that federalism undermines party unity, or that
the availability of a confidence vote mechanism boosts it. There is support,
however, for other hypotheses regarding the effects of competing princi-
pals on unity. When candidates must compete with their own copartisans
for individual support among voters, their responsiveness to personalized
constituencies diminishes party voting unity. Presidents represent another
potential principal to compete with party leaders for legislators’ loyalties,
and cross-voting is more common across the board in systems with elected
presidents. Moreover, although governing parties in parliamentary systems
are highly unified and successful at winning votes, governing parties in
presidential systems have no advantage in voting unity. They are no more
unified or successful on the floor, other things being equal, than opposition
parties. The same distinction applies to governing coalitions in parliamen-
tary versus presidential systems. Whereas the resources associated with
membership in government may be an asset to voting unity under parlia-
mentarism, the potential for presidents to compete with legislative party
and coalition leaders appears to render them a liability to voting unity.
7.3. The Optimal Mix?
7.3.1. The Cases For, and Against, Various Institutional Arrangements

The crux of the competing principals account of party unity advanced here
is that when more than one actor (principal) influences who gets elected
169
under a party label and controls resources legislators care about, divergence
in the demands of these principals will reduce legislative party unity. The
case for collective accountability regards party voting unity as a necessary
condition. I have suggested that institutional arrangements that increase
legislators’ responsiveness to principals other than national party leaders
can push in the direction of individual accountability. The tension between
individual and collective accountability raises the inevitable question of
whether there is some optimal mix of the two, and whether the design of
political institutions can affect whether legislative representation hits
that target.
Academic research on accountability up to now does not provide a
conclusive answer. Some of the most creative recent theoretical work on
accountability focuses on governments or representatives as monolithic
selectors of policy (Manin, Przeworski, and Stokes 1999; Fearon 1999;
Ferejohn 1999), or else on the accountability of presidents alone (Stokes
2001). Among research that discusses legislators and parties explicitly, the
enduring argument for strong-party government holds unity as an unqual-
ified collective good, both in providing coherent options to voters in elec-
tions and in delivering decisive government (American Political Science
Association 1950).
The theme remains central in contemporary scholarship. Powell and
Whitten (1993) include party cohesion as one of the four factors that de-
termine whether voters can assign responsibility to their elected represen-
tatives for policy outcomes, and so hold them accountable. Johnson and
Crisp (2003) demonstrate that the ideological predisposition of legislative
majorities can account for economic policies where president-centered
explanations cannot, but that the connection between legislative party plat-
forms and the policies implemented is stronger when electoral rules dis-
courage individualism. The implication is that collective representation
strengthens the connection between what voters ask for and what they
get. Gerring, Thacker, and Moreno (2005) make a more sweeping claim,
that political institutions that centralize government authority in strong
national parties produce superior policy outcomes on dimensions ranging
from bureaucratic efficiency to investment security to public health and
education. The clear prescription is that representation works best when
legislators answer directly and unequivocally to their parties as principals.
The related theme that institutions that encourage party disunity
produce political pathologies is also prominent. Golden and Chang
(2001) attribute corruption scandals in campaign finance to the degree
170
The Optimal Mix?
of intraparty competition among Christian Democratic legislators in

Italy. Hallerberg and Mairer (2004) contend that personal vote seeking
generates common-pool resource problems whereby legislators under-
value fiscal discipline.
Yet there are competing claims in the scholarship on accountability, with
respect both to centralization of authority and to the idea that legislative
individualism is an unmitigated liability. On centralization, the Madisonian
theme that the division of legislative from executive authority enhances
accountability of representatives retains some support. Persson, Roland,
and Tabellini (1997) argue that competition between the branches, institu-
tionalized under presidentialism, increases the amount of information pol-
iticians supply to citizens about other politicians’ misdeeds, such that the
cumulative effect of increased individual accountability is improved govern-
ment accountability in the aggregate. Hellwig and Samuels (2007) show that
electoral support for presidents’ parties more closely tracks economic growth
rates than does that of prime ministers’ parties in pure parliamentary sys-
tems. It may be that the separation of powers allows for a more specialized
brand of accountability whereby voters can evaluate presidents and legisla-
tors according to their responsiveness to separate sets of demands (Samuels
and Shugart 2003), or that presidentialism’s fixed terms prevent midterm
replacements of the chief executive so common under parliamentarism,
which in turn weaken accountability by sheltering those responsible for
policy failures from voters’ wrath (Cheibub and Przeworski 1999).
Finally, intraparty competition in legislative elections has its own
defenders. Kunicová and Rose-Ackerman (2005) argue that open-list com-
petition discourages collusion among would-be rent seekers, and their data
from across ninety-four countries suggest such elections reduce corrup-
tion. Farrell and McAllister (2004) argue that voters regard the fairness of
elections to be higher and are more satisfied with democracy overall, where
elections allow for personal preference votes among individual legislative
candidates than in closed-list systems.
Current scholarship on democratic institutions provides evidence to
support both the normative goal of collective accountability and the idea
that individual accountability is a democratic asset. Yet the two types of
accountability make contradictory demands on legislators, and we know
relatively little about how the trade-off between these ideals operates. Dis-
cussions of individualism versus collectivism in legislative representation
tend to proceed as though the trade-off were a straightforward matter of
swapping a unit of one sort of accountability for a unit of the other, but
171
accountability is notoriously difficult to measure. And even if we could

measure both individual and collective accountability among legislators,
there is no reason to assume that the substitution of one for the other is
always zero sum.
7.3.2. Individualism versus Individual Accountability

The subtle but critical distinction here is between legislative individualism
and individual accountability. Consider, for example, the model of legisla-
tive individualism proposed by Carey and Shugart (1995), and refined and
applied in various studies since (Wallack, Gaviria, Panizza, and Stein 2003;
Hallerberg and Mairer 2004). The model posits that, in elections with intra-
party competition, incentives for individualism rise monotonically, and
those for collectivism drop correspondingly, as the number of copartisans
against which a given candidate must compete rises. There is empirical
evidence to support this proposition about individualism (Crisp and Ingall
2002; Carey and Reinhardt 2004; Shugart, Valdini, and Suominen 2005).
Yet it does not follow that maximizing individualism per se also maximizes
individual accountability. Accountability rests on the coordination of behav-
ior, information, and sanctioning mechanisms among representatives and
their constituents. Intraparty competition in large multimember districts
might maximize individualism, but larger districts also decrease the quality
of information delivered to voters about specific candidates and increase
coordination problems among voters in sanctioning incumbents (Desposato
2004a; Cox 1997). Recall, for example, the elections in Afghanistan under
the single nontransferable vote (SNTV) rule, described in Chapter 1.
It follows that the conditions to enhance individual accountability are
not necessarily those which maximize individualism and minimize collec-
tive representation. Specifically, individual accountability thrives when
citizens are provided sufficient information about the actions of individual
legislators and are able to use that information to reward or punish at the
polls. When the number of candidates grows too high, the supply and
quality of information about each legislator’s record decline, and the cog-
nitive challenge voters face in processing candidate-specific information
rises (Desposato 2004a; Reynolds 2006). The distinction between legisla-
tive individualism and individual accountability suggests that the tension
between individual and collective accountability might be better repre-
sented as a maximization problem than as an even swap. I defer to future
research the formidable task of measuring the contours of this trade-off
172
The Optimal Mix?
empirically, but the case of Chilean democracy since the reestablishment of

civilian government in 1990 offers an instructive example.
7.3.3. Chile’s Blend of Individualism and Party Unity

In 1989, on the eve of Chile’s transition from military authoritarianism back
to civilian democracy, the outgoing government imposed a unique system
for legislative elections. It called for all elected legislators to be chosen in
open-list competition, but with every district electing two representatives
(i.e., magnitude ¼ 2).1 Votes for both candidates from each electoral alli-
ance, or list, are first pooled in order to determine the distribution of seats
across lists in the district, then candidates from winning lists are awarded
seats in order of their personal preference votes. Thus, the collective per-
formance of the list affects candidates’ prospects, but candidates from the
same list also compete with each other for preference votes.2
Chile’s open-list elections in low-magnitude districts have contributed
to the formation and maintenance of stable multiparty coalitions among
both governing and opposition parties (Siavelis 2000; Londregan 2001).
The coalitions have proved relatively unified in the legislature, to a greater
degree than the ideological proximity of their component parties alone
would predict, and have developed collective reputations that convey sub-
stantial information to voters (Carey 1998). At the same time, Chile’s
combination of presidentialism and personal vote seeking in elections
moderates the tendency toward collectivism in legislative representation.
Figure 7.1 shows WRICEi and RLOSERi indices for parties in the Chilean
Chamber of Deputies, based on recorded votes for a nine-month period at
the end of the second post-transition congress.3 The president’s Christian
1
The outgoing regime’s motives for choosing this system were not benign. It sought to
cushion the anticipated defeat of its civilian politician allies in the ensuing elections. The
unusual conditions by which Chile’s electoral rule was imposed over the objections of its
incipient governing majority mitigate the problem of endogeneity of institutions that is
intractable in much of comparative politics (Przeworski 2007). As a result, Chile presents an
unusually favorable environment for testing the effects of institutions on political behavior.
2
Note that I refer here to electoral alliances as running lists, rather than parties. This is
because Chilean electoral law allows parties to coalesce to run lists, and indeed almost all
seats in Chilean elections under this system have been won by candidates from coalition
lists. As a result, Chilean elections have been characterized by intracoalition competition
but not by intraparty competition.
3
A figure based on votes from the third congress is virtually identical. The mobilization-
based indices, WUNITYi and ULOSER, are suspect in the Chilean case, because the
uniform drop-off across parties from WRICEi to WUNITYi suggests nonvoting equilibria.
173
Figure 7.1 WRICEi and RLOSERi in the Chilean Chamber of Deputies,

1997–98.
Democratic (DC) party is shown at the left of each panel in black with the
other parties in the governing coalition in white, and opposition parties in
varying gray hues toward the right of each panel.
The parties in Chile’s governing coalition experienced more cross-
voting (mean WRICEi ¼ .86) than do governing parties in pure parliamen-
tary systems (mean WRICEi ¼ .93), but less than governing parties in other
presidential systems (mean WRICEi ¼ .76). Chile’s opposition parties were,
174
The Optimal Mix?
on average, just slightly more unified (mean WRICEi ¼ .87) than their
government counterparts. Opposition parties experienced no floor losses
due to cross-voting in this period and governing parties were RLOSERs
on 0.6 percent of votes – the same rate as governing parties in parliamentary
regimes, and below the 3 percent average among governing parties else-
where under presidentialism. These figures suggest sufficient party unity so
that collective representation is viable in Chile, even while individual legis-
lators seek personal votes and voters maintain the ability to retain or reject
specific representatives. In the election following the period on which
Figure 7.1 is based, 85 of 120 Chamber incumbents were nominated for
reelection, and of these, 72 – that is, 85 percent of those on the ballot,
60 percent of all incumbents – won a successive term.
Chilean political institutions are not beyond criticism, and the elec-
toral system, in particular, is subject to regular proposals for reform
(Altman 2005; Huneeus 2006).4 Yet Chilean democracy since 1990 has
delivered a respectable combination of collective and individual legisla-
tive accountability (Cox 2006b). Elections have produced governing coa-
litions that are easily identifiable by voters and that, once in office, have
generally advanced the policies and platforms on which they campaigned.
Governments have met with regular, although not uniform, success on
the floor of Congress. Chilean presidents throughout this period have
occasionally been publicly at odds with the leaders of their allied parties
and coalition, but the more common scenario has been mutual cooperation.
As a result, government-sponsored legislative proposals have mostly been
successful, although sometimes only after prolonged periods of legislative
deliberation (Siavelis 2000; Londregan 2001). The conventional economic
indicators by which governments are judged have been consistently strong in
Chile during this period, and voters have rewarded the governing coalition
with reelection to the presidency three times and returned majorities from
that coalition to the legislature in four consecutive elections. These same
4
The most persistent criticisms stem from the lack of proportionality and the high barriers to
entry inherent in the two-member district system. Increasing district magnitude could rem-
edy this, as opportunities for minority-party representation increase rapidly with increments
in magnitude in this range. In order to improve proportionality while maintaining the con-
ditions for effective individual accountability and avoiding rampant individualism, reformers
might retain candidate preference voting but embrace only modest increases in magnitude.
At magnitudes above 5 or 6, the informational demands on Chilean voters imposed by open-
list competition would threaten individual accountability, and the incentives for individual-
ism could undermine party unity sufficiently to threaten collective accountability.
175
voters can select individual legislators from among fields of candidates small
enough that campaigns are not mere cacophonies of individualistic appeals.
They have taken the opportunity to exercise that discretion, rewarding some,
but not all, incumbents with reelection.
7.3.4. What’s Next?

The search for the optimal balance between collective and individual
accountability among legislators may turn out to be dissatisfying in the
same way as Goldilocks’s method for identifying good porridge. Critics
tend to think they know too much collectivism when they see it, and wide-
spread complaints about ‘‘partyarchy’’ and boss rule suggest they see it
a lot. Other critics see too little collectivism and are prone to complain
about particularism and rudderless parties. There is no mix of individual
and collective accountability widely recognized to be ‘‘just right.’’
This book might represent a step toward identifying that balance, but its
primary goals are descriptive and explanatory, not normative. It describes
two types of accountability and explains how legislative votes can be the
medium through which each is delivered. It documents the levels of trans-
parency and party unity of legislative voting across a wide array of legis-
latures. It offers explanations for why voting transparency has increased in
some instances, and for how and why it can be expected to increase in
others. It explains levels of voting unity according to how the institutional
environment in which legislative parties operate shapes the diversity of
demands placed on lawmakers.
The empirical contribution here is largely one of mapping more of the
legislative world in terms of transparency and party unity. The field of
legislative studies is highly developed in the United States, largely because
the long record of voting transparency in the U.S. Congress has fueled
a vibrant field of study on recorded votes. Transparency is in its relative
infancy in many other legislatures, but has made big strides recently, and
we should expect further advances. This will facilitate mapping the world of
legislative voting more completely and precisely, as well as the develop-
ment of analytical tools that may provide better leverage in evaluating
accountability, and the conditions that enhance and subvert it in its
distinct forms.
176
Appendix: Interview Subjects by Country
Name Office Partya

Bolivia (La Paz, May 14–16, 2001)
Bedregal, Guillermo Deputy MNR
Brockmann, Ericka Senator; party leader MIR
Cárdenas, Vı́ctor Hugo Ex-deputy, ex-vice president MRTA
of the Republic
Carvajal Donoso, Hugo Cabinet minister; MIR
ex-president of Chamber
of Deputies
Ferrufino, Alfonso Ex-deputy MBL
Sánchez de Lozada, Party leader, ex-president of MNR
Gonzalo the Republic
Sánchez Bezraı́n, Carlos Deputy MNR
Colombia (Bogota, May 1–4, 2001)

Acosta, Amilkar Senator PL
Andrade, Hernán Representative PC
Blum de Barberi, Claudia Senador Movimiento ‘98
Devia, Javier Representative PC
Garcı́a Valencia, Jesús Representative PL
Ignacio
Gómez Gallo, Luis Senator PC
Humberto
Guerra, Antonio Senator PL
Gutiérrez, Nancy Patricia Representative PL
Holguı́n Sardi, Carlos Senator PC
Navarro, Antonio Senator MVA
Orduz, Rafael Senator ASI/MCA
Rivera Salazar, Rodrigo Senator PL
(continued)
177
Appendix
Name Office Partya

Costa Rica (San Jose, May 22–26, 2000)
Castillo, Fernando Auditor general of the
Republic
De La Cruz, Vladimir Deputy; party leader PFDN
Gonzalez, Eladio Assembly staff
Guevara, Otto Deputy ML
Guido, Célimo Deputy; party leader PFDN
Hernández, Oscar Assembly staff
Morales, Humberto Assembly staff
Sibaja, Alex Deputy; party leader PLN
Vargas, Eliséo Deputy; party leader PUSC
Vargas Pagán, Carlos Deputy; ex-president of the PUSC
Assembly
Ecuador (Quito, May 18–22, 2001)

Albornoz, Vicente Deputy PS
González, Carlos Deputy ID
Landazuri, Guillermo Deputy ID
Lucero, Wilfredo Deputy; party leader ID
Neira, Xavier Deputy; party leader PSC
Pons, Juan José Deputy DP
Vajello Arcos, Andrés Ex-deputy; ex-president of ID
Congress
Vajello López, Carlos Deputy; ex-president of
Congress
Vela, Alexandra Deputy DP
El Salvador (San Salvador, August 16–18, 2000)

Alvarenga, Aristides Deputy PDC
Alvarenga, Rolando Deputy; party leader ARENA
Duch, Juan Deputy; ex-president of ARENA
Assembly
Pineda, Armando Assembly staff
Zamora, Rubén Ex-deputy; party leader CD
Nicaragua (Managua, August 21–22, 2000)

Baltodano, Mónica Deputy FSLN
Bolaños, Marı́a Lourdes Deputy FSLN
Hurtado, Carlos Deputy; party leader AC
Samper, Jorge Deputy MRS
Urbina Noguera, Luis Deputy PLC
(continued)
178
Appendix
Name Office Partya

Peru (Lima, May 7–9, 2001)
Blanco Oropeza, Carlos Deputy; ex-president of C90-NM
Congress
Cevasco Piedra, José Assembly staff
De Althaus, Jaime Political talk show host
(La Hora N)
Ması́as, Manuel Deputy Independent
Ortiz de Zevallos, Gabriel Pollster (Instituto de Apoyo)
Pease, Henry Deputy; party leader UPP
Venezuela (March 2–10, 2000)

Combellas, Ricardo Deputy (Constituent Independent
Assembly)
Fernández, Julio César Deputy (Interim Assembly) Independent
Murillo, Alexis Assembly staff
Tarek Saab, William Deputy MVR
a
AC: Accion Conservador; ARENA: Alianza para la Renovacion Nacional; ASI/MCA:
Alianza Social Indigena/MCA; C90-NM: Cambio 90 – Nueva Mayoria; CD:
Convergencia Democratica; DP: Democracia y Progreso; FSLN: Frente Sandinista para
la Liberacion Nacional; ID: Izquierda Democratica; MBL: Movimiento Bolivia Libre;
MIR: Movimiento Izquierdista Revolucionario; ML: Movimiento Libertario; MNR:
Movimiento Nacional Revolucionario; MRS: Movimiento Renovacion Sandinismo;
MRTA: Movimiento Revolucionario Tupac Amaru; MVA: Movimiento Via Alterna;
MVR: Movimiento Quinta Republica; PC: Partido Conservador; PDC: Partido
Democrata Cristiana; PFDN: Partido Frente Democratico Nacional; PL: Partido
Liberal; PLC: Partido Liberal Constitucionalista; PLN: Patido Liberacion Nacional; PS:
Partido Socialista; PSC: Partido Social Cristiano; PUSC: Partido Union Social Cristiano;
UPP: Union por el Peru.
179
References
Adserá Alicia, Carles Boix, and Mark Payne. 2003. ‘‘Are You Being Served? Political
Accountability and Quality of Government.’’ Journal of Law, Economics, and
Organization 19(2): 445–90.
Alcántara, Manuel. 1994–2000. Proyecto Elites Latinoamericanos (PELA). Salamanca,
Spain: Instituto Interuniversitario de Iberoamérica (Universidad de Sala-
manca).
Aldrich, John H. 1995. Why Parties? The Origin and Transformation of Political Parties
in America. Chicago: University of Chicago Press.
Aleman, Eduardo, and George Tsebelis. 2005. ‘‘Presidential Conditional Agenda
Setting in Latin America.’’ World Politics 57(3): 396–420.
Altman, David. 2005. ‘‘De un sistema paralelo a un compensatorio (proporcional person-
alizado).’’ Santiago, Chile: Instituto de Ciencia Politica, Pontificia Universidad
Catolica de Chile.
American Political Science Association. 1950. Toward a More Responsible Two-Party
System. New York: Rinehart.
Ames, Barry. 1995. ‘‘Electoral Strategy under Open List Proportional Represen-
tation.’’ American Journal of Political Science 39(2): 406–33.
2002. ‘‘Party Discipline in the Chamber of Deputies.’’ In Legislative Politics in
Latin America, ed. Scott Morgenstern and Benito Nacif, 185–221. Cambridge:
Cambridge University Press.
Amorim Neto, Octavio. 2002. ‘‘Presidential Cabinets, Electoral Cycles, and Coalition
Discipline in Brazil.’’ In Legislative Politics in Latin America, ed. Scott Morgenstern
and Benito Nacif, 48–78. Cambridge: Cambridge University Press.
2006. Presidencialismo e governabilidade nas Americas. Rio de Janeiro: Editora
Fundacion Getulio Vargas.
Amorim Neto, Octavio, Gary W. Cox, and Mathew D. McCubbins. 2003. ‘‘Agenda
Power in Brazil’s Camara dos Deputados, 1989–98.’’ World Politics 55(4):
550–78.
Asamblea Nacional de Panamá. 2005. http://www.asamblea.gob.pa/.
Associación de Derechos Civiles. 2004. ‘‘Ley de Reforma Laboral: Falencias de su
votación.’’ Buenos Aires, Argentina.
181
References
Austen-Smith, David. 1994. ‘‘Strategic Transmission of Costly Information.’’

Econometrica 62: 955–64.
Bagehot, Walter. 1867. The English Constitution. Reprint, Ithaca, NY: Cornell
University Press, 1963.
Baldez, Lisa A., and John M. Carey. 1999. ‘‘Presidential Agenda Control and
Spending Policy: Lessons from General Pinochet’s Constitution.’’ American
Journal of Political Science 43(1): 29–55.
Barczak, Monica. 2001. ‘‘Representation by Consultation? The Rise of Direct
Democracy in Latin America.’’ Latin American Politics and Society 43(3):
37–60.
Bianco, William T., David B. Spence, and John D. Wilkerson. 1996. ‘‘The Elec-
toral Connection in the Early Congress: The Case of the Compensation Act of
1816.’’ American Journal of Political Science 40(1): 145–71.
Bowler, Shaun, David M. Farrell, and Richard S. Katz. 1999. ‘‘Party Cohesion,
Party Discipline, and Parliaments.’’ In Party Discipline and Parliamentary Gov-
ernment, ed. Shaun Bowler, David M. Farrell, and Richard S. Katz, 3–22.
Columbus: Ohio State University Press.
Brady, David W., Joseph Cooper, and Patricia A. Hurley. 1977. ‘‘The Electoral
Basis of Party Voting: Patterns and Trends in the U.S. House of Representa-
tives.’’ In The Impact of the Electoral Process, ed. Louis Misel and Joseph Cooper,
133–65. Beverly Hills: Sage.
Brambor, Thomas, William R. Clark, and Matthew Golder. 2006. ‘‘Understanding
Interaction Models: Improving Empirical Analysis.’’ Political Analysis 14: 63–82.
Brennan, Geoffrey, and Philip Pettit. 1990. ‘‘Unveiling the Vote.’’ British Journal of
Political Science 20(3): 311–33.
Burke, Edmund. 1774. The Works of the Right Honourable Edmund Burke. London:
Henry G. Bohn.
Burns, John F., and Nat Ives. 2005. ‘‘Shiites Win Most Votes in Iraq, Election
Results Show.’’ New York Times, February 13, International sec., 1–22.
Cain, Bruce, John Ferejohn, and Morris Fiorina. 1987. The Personal Vote: Constit-
uency Service and Electoral Independence. Cambridge, MA: Harvard University
Press.
Calvo, Ernesto. 2007. ‘‘The Responsive Legislature: Public Opinion and Discipline
in a Highly Disciplined Legislature.’’ British Journal of Political Science 37:
263–80.
Canes-Wrone, Brandice, David W. Brady, and John F. Cogan. 2002. ‘‘Out of
Step, Out of Office: Electoral Accountability and House Members’ Voting.’’
American Political Science Review 96(1): 127–40.
Carey, John M. 2002. ‘‘Parties and Coalitions in Chile in the 1990s.’’ In Legislative
Politics in Latin America, ed. Scott Morgenstern and Benito Nacif, 222–53.
Cambridge: Cambridge University Press.
2003. ‘‘Discipline, Accountability, and Legislative Voting in Latin America.’’
Comparative Politics 35(2): 191–211.
2007. ‘‘Competing Principals, Political Institutions, and Party Unity in
Legislative Voting.’’ American Journal of Political Science 51(1): 92–107.
182
References
Carey, John M., and Gina Yanitell Reinhardt. 2004. ‘‘State-Level Institutional
Effects on Legislative Coalition Unity in Brazil.’’ Legislative Studies Quarterly
29(1): 23–47.
Carey, John M., and Matthew Soberg Shugart. 1995. ‘‘Incentives to Cultivate
a Personal Vote: A Rank Ordering of Electoral Formulas.’’ Electoral Studies
14(4): 417–39.
1998. ‘‘Calling Out the Tanks, or Filling Out the Forms?’’ In Executive Decree
Authority, ed. John M. Carey and Matthew Soberg Shugart, 1–32. Cambridge:
Carroll, Royce, Gary W. Cox, and Monica Pachon. 2006. ‘‘How Political Parties
Create Democracy, Chapter 2.’’ Legislative Studies Quarterly 31(2): 153–74.
Centro de Estudios Públicos. 2007. Estudio nacional de opinion pública. Santiago,
Chile: Centro de Estudios Públicos.
Cevasco Piedra, José. 2000. El Congreso del Perú: Un modelo de modernizacioń. Lima:
Ediciones del Congreso del Perú.
Cheibub, Jose Antonio, and Adam Przeworski. 1999. ‘‘Democracy, Elections, and
Accountability for Economic Outcomes.’’ In Democracy, Accountability, and
Representation, ed. Bernard Manin, Adam Przeworski, and Susan C. Stokes,
222–50. Cambridge: Cambridge University Press.
Cheibub, José Antonio, Adam Przeworski, and Sebastian M. Saiegh. 2004.
‘‘Government Coalitions and Legislative Effectiveness under Presidentialism
and Parliamentarism.’’ British Journal of Political Science 34: 565–87.
Chhibber, Pradeep, and Ken Kollman. 1998. ‘‘Party Aggregation and the Number
of Parties in India and the United States.’’ American Political Science Review
92(2): 329–42.
Clarıń. 2004a. ‘‘La justicia resolvera si debe quedar registrado como vota cada
diputado porteno.’’ Buenos Aires, Argentina. February 17. www.clarin.com.
2004b. ‘‘Otra forma de votar las leyes.’’ Buenos Aires, Argentina. February 18.
www.clarin.com.
Cleary, Matthew R., and Susan C. Stokes. 2006. Democracy and the Culture of Skep-
ticism: Political Trust in Argentina and Mexico. New York: Russell Sage Foun-
dation.
Colomer, Josep M. 2001. Political Institutions: Democracy and Social Choice. New
York: Oxford University Press.
Condorcet, Marquis de. 1785. Essay on the Application of Analysis to the Probability of
Majority Decisions (Essai sur l’application de l’analyse à la probabilté des deci-
sions rendues a la pluralité des voix). Paris: De l’Imprimerie royale.
Congreso del Perú. 1998. Diario de los debates. Primera Legislatura Ordinaria de
1998, 11a Sesión, 24 de setiembre.
Coppedge, Michael J. 1994. Strong Parties and Lame Ducks: Presidentialism,
Partyarchy, and Factionalism in Venezuela. Stanford: Stanford University Press.
1998. ‘‘The Dynamic Diversity of Latin American Party Systems.’’ Party Politics
4(4): 547–68.
Corrado, Anthony. 2002. ‘‘A History of Federal Campaign Finance Law.’’ In The
New Campaign Finance Sourcebook, ed. Anthony Corrado, Thomas Mann,
183
References
Daniel Ortiz, and Trevor Potter. Washington, DC: Brookings Institution.

http://www.brook.edu/dybdocroot/gs/cf/sourcebk01/HistoryChap.pdf.
Cox, Gary W. 1987. The Efficient Secret. Cambridge: Cambridge University Press.
1990. ‘‘Centripetal and Centrifugal Incentives in Electoral Systems.’’ American
1997. Making Votes Count: Strategic Coordination in the World’s Electoral Systems.
2006a. ‘‘The Organization of Democratic Legislatures.’’ In Oxford Handbook of
Political Economy, ed. Barry Weingast and Donald Wittman, 141–61. New
2006b. ‘‘Evaluating Electoral Systems.’’ Revista de Ciencia Politica 26(1): 212–15.
Cox, Gary W., and Mathew D. McCubbins. 1993. Legislative Leviathan: Party
Government in the House. Berkeley: University of California Press.
2005. Setting the Agenda: Responsible Party Government in the U.S. House of
Representatives. Cambridge: Cambridge University Press.
Cox, Gary W., Mikitaka Masuyama, and Mathew D. McCubbins. 2000. ‘‘Agenda
Power in the Japanese House of Representatives.’’ Japanese Journal of Political
Science 1(1): 1–12.
Cox, Gary W., and Keith T. Poole. 2004. ‘‘On Measuring Partisanship in Roll-Call
Voting: The US House of Representatives, 1877–1999.’’ American Journal of
Cox, Gary W., and Matthew S. Shugart. 1995. ‘‘In the Absence of Vote Pooling:
Nomination and Vote Allocation Errors in Colombia.’’ Electoral Studies 14(4):
441–60.
Cox, Gary W., and Michael F. Thies. 1998. ‘‘The Cost of Intraparty Competition:
The Single, Nontransferable Vote and Money Politics in Japan.’’ Comparative
Political Studies 31(3): 267–91.
Crisp, Brian F., and Rachel E. Ingall. 2002. ‘‘Institutional Engineering and the
Nature of Representation: Mapping the Effects of Electoral Reform in Colom-
bia.’’ American Journal of Political Science 46(4): 733–48.
Crisp, Brian, and Juan Carlos Rey. 2001. ‘‘The Causes of Electoral Reform in
Venezuela.’’ In Mixed-Member Electoral Systems: The Best of Both Worlds, ed.
Matthew S. Shugart and Martin P. Wattenberg, 173–93. New York: Oxford
University Press.
Culver, William W., and Alfonso Ferrufino. 2000. ‘‘Diputados uninominales: La
participacion politica en Bolivia.’’ Contribuciones 1: 1–28.
Dallas Morning News. 2003a.‘‘Editorial: Let the Sun Shine – Legislature Needs to
Get Online with Records.’’ Dallas Morning News Online, April 1. http://
www.dallasnews.com/.
2003b.‘‘Editorial: Let the Sun Shine – Excuses for Not Recording Votes Are
Flimsy.’’ Dallas Morning News Online, March 31. http://www.dallasnews.com/.
Dawisha, Adeed, and Larry Diamond. 2006. ‘‘Iraq’s Year of Voting Dangerously.’’
Journal of Democracy 17(2): 89–103.
Desposato, Scott. 2004a. ‘‘Going Negative in Comparative Perspective: The
Impact of Electoral Rules on Campaign Strategy.’’ Paper presented at the
184
References
annual meeting of the American Political Science Association, Boston, August

30–September 3.
2004b. ‘‘The Impact of Federalism on National Party Cohesion in Brazil.’’
Legislative Studies Quarterly 29(2): 259–85.
2005. ‘‘Correcting for Small-Group Cohesion in Roll Call Cohesion Scores.’’
British Journal of Political Science 35(4): 731–44.
2006a. ‘‘Parties for Rent? Ambition, Ideology, and Party-Switching in Brazil’s
Chamber of Deputies.’’ American Journal of Political Science 50(1): 62–80.
2006b. ‘‘Brazilian Political Reform: What Needs Fixing and How to Fix It.’’
Paper presented at the Center for Brazilian Studies, Oxford University, June 5.
Diermeier, Daniel, and Timothy J. Feddersen. 1998. ‘‘Cohesion in Legislatures and
the Vote of Confidence Procedure.’’ American Political Science Review 92(3):
611–22.
Diez Canseco, Javier. 2001. ‘‘Impondra el Congreso mano blanda y lucro cesante?’’
La Republica, May 6, p. 24.
Duverger, Maurice. 1954. Political Parties, Their Organization and Activity in the
Modern State. New York: Wiley.
El Tiempo. 2002. ‘‘Alvaro Uribe Velez radico proyecto de referendo en su primer
acto de gobierno.’’ August 7. http://www.eltiempo.com/.
Erickson, Robert S. 1971. ‘‘The Electoral Impact of Congressional Roll Call
Voting.’’ American Political Science Review 65: 1018–32.
Farrell, David M., and Ian McAllister. 2004. ‘‘Voter Satisfaction and Electoral
Systems: Does Preferential Voting in Candidate-Centered Systems Make
a Difference?’’ Paper presented at the University of California, Irvine. Posted
in eScholarship Repository: http://repositories.cdlib.org/csd/04-04.
Fearon, James D. 1999. ‘‘Electoral Accountability and the Control of Politicians:
Selecting Good Types versus Sanctioning Poor Performance.’’ In Democracy,
Accountability, and Representation, ed. Bernard Manin, Adam Przeworski, and
Susan C. Stokes, 55–97. Cambridge: Cambridge University Press.
Fenno, Richard F., Jr. 1978. Home Style: House Members in Their Districts. Boston:
Little, Brown.
Ferejohn, John. 1999. ‘‘Accountability and Authority: Toward a Theory of Political
Accountability.’’ In Democracy, Accountability, and Representation, ed. Bernard
Manin, Adam Przeworski, and Susan C. Stokes, 131–53. Cambridge:
Figueiredo, Argelina Cheibub, and Fernando Limongi. 2000. ‘‘Presidential Power,
Legislative Organization, and Party Behavior in Brazil.’’ Comparative Politics
32(2): 151–70.
Fleischer, David. 2004. ‘‘The Politics of the Lula PT-led Coalition.’’ Paper pre-
pared for the conference of the Latin American Studies Association, Las Vegas,
NV, November 4–6.
Frye, Timothy. 1997. ‘‘A Politics of Institutional Choice: Post-Communist
Presidencies.’’ Comparative Political Studies 30(5): 523–52.
Gallagher, Michael. 1991. ‘‘Proportionality, Disproportionality and Electoral
Systems.’’ Electoral Studies 10: 33–51.
185
References
Garman, Christopher, Stephan Haggard, and Eliza Willis. 2001. ‘‘Fiscal Decen-
tralization: A Political Theory with Latin American Cases.’’ World Politics
53(2): 205–36.
Gerring, John, and Strom C. Thacker. 2008. A Centripetal Theory of Democratic
Governance. Cambridge: Cambridge University Press.
Gerring, John, Strom C. Thacker, and C. Moreno. 2005. ‘‘Centripetal Democratic
Governance: A Theory and Global Inquiry.’’ American Political Science Review
99(4): 567–81.
Golden, Miriam A., and Eric C.C. Chang. 2001. ‘‘Competitive Corruption:
Factional Conflict and Political Malfeasance in Postwar Italian Christian
Democracy.’’ World Politics 53(4): 588–622. http://muse.jhu.edu/journals/
world_politics/ v053/53.4golden.pdf.
Gonzalez Bertomeu, Juan F. 2004. Cada voto con su nombre: Votaciones nominales en el
Poder Legislativo. Buenos Aires: Asociación por los Derechos Civiles.
Gunson, Phil. 2006. ‘‘Chavez’s Venezuela.’’ Current History 105(688): 58–63.
Hallerberg, Mark, and Patrik Mairer. 2004. ‘‘Executive Authority, the Personal
Vote, and Budget Discipline in Latin American and Caribbean Countries.’’
American Journal of Political Science 48(3): 571–87.
Haspel, Moshe, Thomas Remington, and Steven S. Smith. 1998. ‘‘Electoral Insti-
tutions and Party Cohesion in the Russian Duma.’’ Journal of Politics 60(2):
417–39.
Hellwig, Timothy, and David Samuels. 2007. ‘‘Electoral Accountability and the
Variety of Democratic Regimes.’’ British Journal of Political Science 38: 65–90.
Hix, Simon. 2002. ‘‘Parliamentary Behavior with Two Principals: Preferences,
Parties, and Voting in the European Parliament.’’ American Journal of Political
Science 46(3): 688–98.
2004. ‘‘Electoral Institutions and Legislative Behavior – Explaining Voting
Defection in the European Parliament.’’ World Politics 56(2): 194–223.
Hix, Simon, Abdul Noury, and Gerard Roland. 2006. ‘‘Dimensions of Politics in
the European Parliament.’’ American Journal of Political Science 50(2): 494–511.
Huber, John D. 1996. ‘‘The Vote of Confidence in Parliamentary Democracies.’’
American Political Science Review 90(2): 269–82.
Huber, John D., and G. Bingham Powell. 1994. ‘‘Congruence between Citizens
and Policymakers in Two Visions of Liberal Democracy.’’ World Politics 46:
291–326.
Huneeus, Carlos. ‘‘Binominalismo y calidad democratica.’’ La Tercera, Santiago,
Chile, August 8, 2006, Opinion sec.
International Institute for Democracy and Electoral Assistance. 1997. Handbook of
Electoral Systems. Stockholm, Sweden: IDEA.
Jenkins, Jeffrey A. 1999. ‘‘Examining the Bonding Effects of Party: A Comparative
Analysis of Roll-Call Voting in the US and Confederate Houses.’’ American
2000. ‘‘Examining the Robustness of Ideological Voting: Evidence from the
Confederate House of Representatives.’’ American Journal of Political Science
44(4): 811–22.
186
References
Jenkins, Jeffrey A., and Charles Stewart III. 2003. ‘‘Out in the Open: The Emer-
gence of Viva Voce Voting in House Speakership Elections.’’ Legislative Studies
Quarterly 28(4): 481–508.
Johnson, Chris, William Maley, Alexander Thier, and Ali Wardak. 2003. Afghani-
stan’s Political and Constitutional Development. London: Overseas Development
Institute and U.K. Department for International Development.
Johnson, Gregg, and Brian F. Crisp. 2003. ‘‘Mandates, Powers, and Policies.’’
Jones, Mark P., and W. Hwang. 2005. ‘‘Party Government in Presidential Democ-
racies: Extending Cartel Theory beyond the US Congress.’’ American Journal
of Political Science 49(2): 267–82.
Katz, Richard S. 1994. ‘‘The 1993 Parliamentary Electoral Reform.’’ In Italian
Politics, ed. Carol Mershon and Gianfranco Pasquino, 93–112. 9th ed. Boulder,
CO: Westview Press.
Kile, Orville M. 1948. The Farm Bureau through Three Decades. Baltimore: Waverly
Press.
King, David C., and Richard J. Zeckhauser. 2003. ‘‘Congressional Vote Options.’’
Krehbiel, Keith. 1998. Pivotal Politics: A Theory of U.S. Lawmaking. Chicago:
University of Chicago Press.
2000. ‘‘Party Discipline and Measures of Partisanship.’’ American Journal of
Kulischeck, Michael, and Brian F. Crisp. 2001. ‘‘The Consequences of Electoral
Reform in Venezuela.’’ In Mixed-Member Electoral Systems: The Best of Both
Worlds?, ed. Matthew S. Shugart and Martin P. Wattenberg, 404–31. New
Kunicová, Jana, and Susan Rose-Ackerman. 2005. ‘‘Electoral Rules and Constitu-
tional Structures as Constraints on Corruption.’’ British Journal of Political
Science 35(4): 573–606.
La Gaceta. 2001. ‘‘Congresistas que votaron y en contra de la revocatoria.’’ La
Gaceta, 6 May, sec. 2.
La Raja, Raymond J. 2003. ‘‘Why Soft Money Has Strengthened Parties.’’ In Inside
the Campaign Finance Battle: Court Testimony on the New Reforms, ed. Anthony
Corrado, Thomas E. Mann, and Trevor Potter, 69–98. Washington, DC:
Brookings Institution Press.
Laakso, Markku, and Rein Taagepera. 1979. ‘‘Effective Number of Parties: A Mea-
sure with Application to West Europe.’’ Comparative Political Studies 12: 3–27.
Lanfranchi, Prisca, and Ruth Luthi. 1999. ‘‘Cohesion of Party Groups and Inter-
party Conflict in the Swiss Parliament: Roll Call Voting in the National
Council.’’ In Party Discipline and Parliamentary Government, ed. Shaun Bowler,
David M. Farrell, and Richard S. Katz, 99–120. Columbus: Ohio State
University Press.
Latin America Data Base. 2001. ‘‘Nicaragua: Governing Party and Sandinistas
Nominate Presidential Candidates.’’ NotiCen: Central American & Caribbean
Political & Economic Affairs 6(3).
187
References
Latinobarómetro. 2003 Summary Report: Democracy and Economy. www

.latinobarometro.org.
Laver, Michael, and Kenneth Shepsle. 1996. Making and Breaking Governments:
Cabinets and Legislatures in Parliamentary Democracies. Cambridge: Cambridge
University Press.
Lijphart, Arend. 1994. Electoral Systems and Party Systems: A Study of 27 Democracies,
1945–1990. New York: Oxford University Press.
1999. Patterns of Democracy: Government Forms and Performance in Thirty-Six
Countries. New Haven: Yale University Press.
Linz, Juan J. 1994. ‘‘Presidentialism or Parliamentarism: Does It Make a Differ-
ence?’’ In The Failure of Presidential Democracy, ed. Juan J. Linz and Arturo
Valenzuela, 3–90. Baltimore: Johns Hopkins University Press.
Londregan, John. 2001. Legislative Institutions and Ideology in Chile. Cambridge:
Lowry, W. R., and C. R. Shipan. 2002. ‘‘Party Differentiation in Congress.’’
Luna, Juan P., and Elizabeth J. Zechmeister. 2005. ‘‘Political Representation
in Latin America: A Study of Elite-Mass Congruence in Nine Countries.’’
Comparative Political Studies 38(4): 388–416.
Lupia, Arthur, and Mathew D. McCubbins. 1998. The Democratic Dilemma: Can
Citizens Learn What They Need to Know? Cambridge: Cambridge University Press.
Mainwaring, Scott P. 1999. Rethinking Party Systems in the Third Wave of Democra-
tization: The Case of Brazil. Stanford: Stanford University Press.
Mainwaring, Scott P., and Anı́bal Pérez-Liñán. 1997. ‘‘Party Discipline in
the Brazilian Constitutional Congress.’’ Legislative Studies Quarterly 22(4):
453–83.
Mainwaring, Scott P., and Timothy R. Scully. 1995. ‘‘Party Systems in Latin
America.’’ In Building Democratic Institutions: Party Systems in Latin America,
ed. Scott Mainwaring and Timothy R. Scully, 1–36. Stanford: Stanford
University Press.
Manin, Bernard, Adam Przeworski, and Susan C. Stokes. 1999. ‘‘Elections and
Representation.’’ In Elections, Accountability, and Representation, ed. Bernard
Manin, Adam Przeworski, and Susan C. Stokes, 29–54. Cambridge:
Mann, Thomas E. 2003. ‘‘Linking Knowledge and Action: Political Science and
Campaign Finance Reform.’’ Perspective on Politics 17(1): 69–83.
Marlowe, Ann. 2007. ‘‘Life of the Parties.’’ Wall Street Journal, January 29, A17.
Mayhew, David. 1974. Congress: The Electoral Connection. New Haven: Yale
University Press.
Mayorga, Rene. 2001. ‘‘Electoral Reform in Bolivia: Origins of the Mixed-Member
Proportional System.’’ In Mixed-Member Electoral Systems: The Best of Both
Worlds?, ed. Matthew S. Shugart and Martin P. Wattenberg, 194–208. New
McCoy, J. L. 1999. ‘‘Chavez and the End of ‘Partyarchy’ in Venezuela.’’ Journal of
Democracy 105(3): 64–77.
188
References
McCubbins, Mathew D., and Frances McCall Rosenbluth. 1995. ‘‘Party Provision
for Personal Politics: Dividing the Vote in Japan.’’ In Structure and Policy in
Japan and the United States, ed. Peter F. Cowhey and Mathew D. McCubbins,
35–55. Cambridge: Cambridge University Press.
McKelvey, Richard. 1976. ‘‘Intransitivities in Multidimensional Voting Models:
Some Implications for Agenda Control.’’ Journal of Economic Theory 12: 472–82.
Moe, Terry M., and Michael Caldwell. 1994. ‘‘The Institutional Foundations of
Democratic Government: A Comparison of Presidential and Parliamentary
Systems.’’ Journal of Institutional Theoretical Economics 150(1): 171–95.
Molinar Horcacitas, Juan, and Jeffrey Weldon. 2001. ‘‘Reforming Electoral Sys-
tems in Mexico.’’ In Mixed-Member Electoral Systems: The Best of Both Worlds?,
ed. Matthew Soberg Shugart and Martin P. Wattenberg, 209–30. Oxford:
Oxford University Press.
Morgan, Jana, Rosario Espinal, and Mitchell A. Seligson. 2006. The Political Culture
of the Dominican Republic in 2006. Nashville, TN: Latin American Public Opin-
ion Project.
Morgenstern, Scott. 2003. Patterns of Legislative Politics: Roll Call Voting in Latin
America and the United States. Cambridge: Cambridge University Press.
Morone, James A. 1990. The Democratic Wish: Popular Participation and the Limits of
American Government. New York: Basic Books.
Nicolau, Jairo Marconi. 2007. ‘‘O Sistema Eleitoral De Lista Aberta no Brasil.’’ In
Instituicxo˜es representativas no Brasil: Balancxo e reforma, ed. Jairo Marconi Nicolau
and Timothy J. Power, 97–122. Rio de Janeiro: IUPERJ.
Odegard, Peter H. 1928. Pressure Politics: The Story of the Anti-Saloon League. New
York: Columbia University Press.
Pérez-Liñán, Anı́bal. 2007. Presidential Impeachment and the New Political Instability
in Latin America. Cambridge: Cambridge University Press.
Persson, Torsten, Gerard Roland, and Guido Tabellini. 1997. ‘‘Separation of
Powers and Political Accountability.’’ Quarterly Journal of Economics 112:
1163–1202.
Persson, Torsten, and Guido Tabellini. 2003. The Economic Effects of Constitutions.
Cambridge, MA: MIT Press.
Persson, Torsten, Guido Tabellini, and F. Trebbi. 2003. ‘‘Electoral Rules and
Corruption.’’ Journal of the European Economic Association 1: 958–89.
Pharr, Susan J., Robert D. Putnam, and Russell J. Dalton. 2000. ‘‘A Quarter-
Century of Declining Confidence.’’ Journal of Democracy 11(2): 500–25.
Poole, Keith T. 1998. ‘‘Recovering a Basic Space from a Set of Issue Scales.’’
Poole, Keith T., and Howard Rosenthal. 1985. ‘‘A Spatial Model for Legislative
Roll Call Analysis.’’ American Journal of Political Science 29(2): 357–84.
1997. Congress: A Political-Economic History of Roll Call Voting. New York: Oxford
University Press.
2001. ‘‘D-Nominate after 10 Years: A Comparative Update to Congress; A
Political-Economic History of Roll-Call Voting.’’ Legislative Studies Quarterly
26(1): 5–30.
189
References
Powell, G. Bingham. 2000. Elections as Instruments of Democracy: Majoritarian and

Proportional Visions. New Haven: Yale University Press.
Powell, G. Bingham, and Georg Vanberg. 2000. ‘‘Election Laws, Dispro-
portionality and Median Correspondence: Implications for Two Visions of
Democracy.’’ British Journal of Political Science 30(3): 383–411. http://www
.journals.cup.org/bin/bladerunner?REQUNIQ=1028572637&;;REQSESS=
75269&117000REQEVENT=&REQINT1=1202&REQAUTH=0.
Powell, G. Bingham, and Guy D. Whitten. 1993. ‘‘A Cross-National Analysis of
Economic Voting: Taking Account of the Political Context.’’ American Journal
of Political Science 37(2): 391–414.
Proyecto Élites Parlamentarias Latinoamericanas. 2006. Boletıń n°6: Disciplina de
voto y lealtad partidista. Salamanca, Spain: Instituto Interuniversitario de Iber-
oamérica (Universidad de Salamanca).
Przeworski, Adam. 2007. ‘‘Is the Science of Comparative Politics Possible?’’ In
Oxford Handbook of Comparative Politics, ed. Carles Boix and Susan C. Stokes,
117–39. New York: Oxford University Press.
Quinn, Kevin M., Burt L. Monroe, Michael Colaresi, Michael H. Crespin,
and Dragomir R. Radev. 2006. ‘‘An Automated Method of Topic-Coding
Legislative Speech over Time with Application to the 105th–108th U.S.
Senate.’’ Paper presented at the annual meetings of the American Political
Science Association, Philadelphia, September 2.
Rachadell, Manuel. 1991. ‘‘El sistema electoral y la reforma de los partidos.’’
In Venezuela, democracia y futuro: Los partidos politicos en la decada de los
90. Reflexiones para un cambio neceario, ed. Carlos Blanco and Edgar
Paredes Pisani, 203–10. Caracas: Comision Presidencial para la Reforma del
Estado.
Rasch, Bjorn E. 2000. ‘‘Parliamentary Floor Voting Procedures and Agenda Setting
in Europe.’’ Legislative Studies Quarterly 25(1): 3–23.
Reynolds, Andrew. 2006. ‘‘The Curious Case of Afghanistan.’’ Journal of Democracy
17(2): 104–17.
Rice, Stuart A. 1925. ‘‘The Behavior of Legislative Groups.’’ Political Science
Quarterly 40: 60–72.
Riker, William H. 1982. Liberalism against Populism. Prospect Heights, IL:
Waveland Press.
Roberts, Kenneth, and Erik Wibbels. 1999. ‘‘Party Systems and Electoral Volatility
in Latin America: A Test of Economic, Institutional, and Structural Explan-
ations.’’ American Political Science Review 93(3): 575–90.
Rosas, Guillermo. 2005. ‘‘The Ideological Organization of Latin American
Legislative Parties: An Empirical Analysis of Elite Policy Preferences.’’
Comparative Political Studies 7(824): 849.
Rose-Ackerman, Susan. 1999. Corruption and Government: Causes, Consequences, and
Reform. Cambridge: Cambridge University Press.
Rosenthal, Howard, and Erik Voeten. 2004. ‘‘Analyzing Roll Calls with Perfect
Spatial Voting: France 1946–1958.’’ American Journal of Political Science 48(3):
620–32.
190
References
Rousseau, Jean Jacques. 1763. ‘‘Voting.’’ In On the Social Contract, book IV, ch. XX.
New York: St. Martin’s Press, 1978.
Rubin, Barnett R. ‘‘The Wrong Voting System.’’ International Herald Tribune,
March 16, International section.
Samuels, David. 2000. ‘‘Ambition and Competition: Explaining Legislative Turn-
over in Brazil.’’ Legislative Studies Quarterly 25(3): 481–97.
2004. ‘‘Presidentialism and Accountability for the Economy in Comparative
Perspective.’’ American Political Science Review 98(3): 425–36.
Samuels, David, and Matthew S. Shugart. 2003. ‘‘Presidentialism, Elections, and
Representation.’’ Journal of Theoretical Politics 15(1): 33–60.
Sartori, Giovanni. 1976. Parties and Party Systems: A Framework for Analysis.
Scarrow, Susan E. 2001a. ‘‘Direct Democracy and Institutional Change: A
Comparative Investigation.’’ Comparative Political Studies 34(688): 651–65.
2001b. ‘‘Germany: The Mixed-Member System as a Political Compromise.’’
In Mixed-Member Electoral Systems: The Best of Both Worlds?, ed. Matthew
S. Shugart and Martin P. Wattenberg, 55–69. New York: Oxford University
Press.
Seligson, Mitchell A., Abby B. Cordova, Juan Carlos Donoso, Daniel Moreno
Morales, Diana Orces, and Vivian Schwarz Blum. 2007. Democracy Audit:
Bolivia 2006 Report. Nashville, TN: Latin American Public Opinion Project.
Shepsle, Kenneth. 1991. Models of Multiparty Electoral Competition. New York:
Harwood Academic.
Shugart, Matthew S. 1998. ‘‘The Inverse Relationship between Party Strength and
Executive Strength: A Theory of Politicians’ Constitutional Choices.’’ British
Shugart, Matthew S., and John M. Carey. 1992. Presidents and Assemblies:
Constitutional Design and Electoral Dynamics. Cambridge: Cambridge University
Press.
Shugart, Matthew S., Erika Moreno, and Luis E. Fajardo. 2006. ‘‘Deepening
Democracy by Renovating Political Practices: The Struggle for Electoral
Reform in Colombia.’’ In Peace, Democracy, and Human Rights in Colombia,
ed. Christopher Welna and Gustavo Gallon, 202–66. Notre Dame, IN: Notre
Dame University Press.
Shugart, Matthew S., Melody E. Valdini, and K. Suominen. 2005. ‘‘Looking for
Locals: Voter Information Demands and Personal Vote-Earning Attributes of
Legislators under Proportional Representation.’’ American Journal of Political
Science 49(2): 437–49.
Shugart, Matthew S., and Martin P. Wattenberg, eds. 2001. Mixed-Member Electoral
Systems: The Best of Both Worlds? New York: Oxford University Press.
Siavelis, Peter. 2000. The President and Congress in Postauthoritarian Chile: Institu-
tional Constraints to Democratic Consolidation. University Park: Pennsylvania
State University Press.
Sieberer, Ulrich. 2006. ‘‘Party Unity in Parliamentary Democracies: A Compara-
tive Analysis.’’ Journal of Legislative Studies 12(2): 150–78.
191
References
Singer, Judith D. 1998. ‘‘Using SAS PROC MIXED to Fit Multilevel Models,
Hierarchical Models, and Individual Growth Models.’’ Journal of Educational
and Behavioral Statistics 24(4): 323–55.
Skeen, C. Edward. 1986. ‘‘Vox Populi, Vox Dei: The Compensation Act of 1816
and the Rise of Popular Politics.’’ Journal of the Early Republic 6(253): 274.
Smith, Steven S. 1989. Call to Order: Floor Politics in the House and Senate.
Washington, DC: Brookings Institution Press.
Snyder, James M., and Tim Groseclose. 2000. ‘‘Estimating Party Influence in
Congressional Roll-Call Voting.’’ American Journal of Political Science 44(2):
193–211.
Snyder, James M., and Michael M. Ting. 2005. ‘‘Why Roll Calls? A Model of
Position Taking in Legislative Voting and Elections.’’ Journal of Law, Econom-
ics, and Organization 21(1): 153–78.
Steenbergen, Marco R., and Bradford S. Jones. 2002. ‘‘Modeling Multilevel Data
Structures.’’ American Journal of Political Science 46(1): 218–237.
Stokes, Susan C. 2001. Mandates and Democracy: Neoliberalism by Surprise in Latin
America. Cambridge: Cambridge University Press.
Sundquist, James L. 1981. The Decline and Resurgence of Congress. Washington, DC:
Brookings Institution Press.
Taagepera, Rein, and Matthew Sobert Shugart. 1989. Seats and Votes: The Effects and
Determinants of Electoral Systems. New Haven: Yale University Press.
Tapia G., Sady. 2004. ‘‘Legisladores ‘huyen’ del voto electrónico.’’ La Prensa,
March 25. La Prensa Web, http://www.prensa.com/.
Texas State Legislature. 2005. ‘‘Texas Legislature Online.’’ http://www.capitol
.state.tx.us/tlo/legislation/voteinfo.htm.
Thames, Francis C. 2007. ‘‘Searching for the Electoral Connection: Parliamentary
Party Switching in the Ukrainian Rada, 1998–2002.’’ Legislative Studies Quar-
terly 32(2): 223–56.
Tsebelis, George. 1999. ‘‘Veto Players and Law Production in Parliamentary
Democracies: An Empirical Analysis.’’ American Political Science Review 93(3):
591–608.
Ungar Bleier, Elisabeth. 2002. ‘‘Consideraciones sobre la reforma politica.’’
Testimony before the Comisioń Primera Constitucional del Senado de la República.
Bogota´.
United States Supreme Court. 1958. Majority Decision in NAACP v. Alabama
Ex Rel. Patterson, Attorney General. No. 91 Supreme Court of the United States
357 U.S. 449; 78 S. Ct. 1163.
Valenzuela, Arturo. 1994. ‘‘Party Politics and the Crisis of Presidentialism in
Chile: A Proposal for a Parliamentary Form of Government.’’ In The Failure
of Presidential Democracy: The Case of Latin America, vol. 2, ed. Juan J. Linz and
Arturo Valenzuela, 91–150. Baltimore: Johns Hopkins University Press.
VoteWorld. 2008. VoteWorld: The International Legislative Roll-Call Voting Website.
http://ucdata.berkeley.edu:7101/new_web/VoteWorld/voteworld/.
Wallack, Jessica S., A. Gaviria, Ugo Panizza, and Eduardo Stein. 2003. ‘‘Particu-
larism around the World.’’ World Bank Economic Review 17(1): 133–43.
192
References
Weldon, Jeffrey. 1997. ‘‘Political Sources of Presidencialismo in Mexico.’’ In Presi-

dentialism and Democracy in Latin America, ed. Matthew S. Shugart and Scott P.
Mainwaring, 225–58. Cambridge: Cambridge University Press.
2001. ‘‘The Consequences of Mexico’s Mixed-Member Electoral System,
1988–1997.’’ In Mixed-Member Electoral Systems: The Best of Both Worlds?, ed.
Matthew S. Shugart and Martin P. Wattenberg, 447–74. New York: Oxford
University Press.
Weyland, Kurt. 1996. Democracy without Equity: Failures of Reform in Brazil.
Pittsburgh: University of Pittsburgh Press.
White, Leonard D. 1951. The Jeffersonians: A Study in Administrative History,
1801–1829. New York: Macmillan.
Whiteley, Paul F., and Patrick Seyd. 1999. ‘‘Discipline in the British Conservative
Party: The Attitudes of Party Activists toward the Role of Their Members
of Parliament.’’ In Party Discipline and Parliamentary Government, ed. Shaun
Bowler, David M. Farrell, and Richard S. Katz, 53–71. Columbus: Ohio State
University Press.
193
Index
accountability, 1–4, 7–9, 11, 16, 29–36, Brazil, 15, 18, 25, 45, 47, 58, 60, 61, 71,
43, 45, 46, 51–55, 63, 64, 65 75, 88, 105, 108, 110, 125, 141,
68–70, 74, 80, 82, 85, 89–91 142, 144, 155, 156, 157, 163
92, 93, 165–172, 175, 176 Burke, Edmund, 69, 70, 91
collective, 1, 2, 9, 10, 12–14, 20–21,
24, 29, 92, 166, 168–176 Calvo, Ernesto, 47, 135
individual, 1–3, 9, 12, 13, 15–16, 20, campaign contributions, 85
21, 32, 36, 41, 44, 166–168 Canada, 103, 104, 105, 107, 108, 141,
170–172 142, 154, 155
rendiciones de cuentas, 33 Carazo Odio, Rodrigo, 79
Afghanistan, 9–13, 172 Cardoso, Fernando Henrique, 110,
agenda powers, 5, 17, 18, 25, 44, 48, 76, 155, 156
77, 78, 81, 91, 125–128, 132 Chavez, Hugo, 27, 32, 33, 34, 90, 137
134–136, 166 Chile, 37, 45, 58, 60, 61, 71, 88, 104,
American Legion, 54 105, 107, 108, 140, 141, 173
American Medical Association, 54 174, 175
American Political Science Association, CLOSE score, 104, 116
4, 23, 170 coalitions, 6, 15, 16, 21, 22, 29, 47, 48,
Ames, Barry, 47, 97, 125 53, 64, 65, 75–77, 78, 80, 88, 90,
Antigone. See Sophocles 92, 93, 99, 109, 110, 113, 126–127,
Anti-Saloon League, 54 135, 155–163, 166–169, 173
Argentina, 15, 45, 50, 58, 59, 60, 61, 64, 174–176
71, 75, 81, 90, 98, 99, 105, 107, governing, 159
108, 140, 141, 142 opposition, 159, 161
Austen-Smith, David, 33 cohesiveness, 7, 27, 123, 126–129
Australia, 103, 105, 107, 108, 109, 141 132, 162
Coleman, William, 52
Bagehot, Walter, 23, 139, 141 Colombia, 25, 26, 27, 38–41, 45, 49, 57,
Baltodano, Mónica, 28, 29, 80, 89, 136 58, 60, 61, 64, 66, 71, 78, 79
Blanco Oropeza, Carlos, 84, 89, 91 80–84, 90
Bolivia, 25, 26, 28, 31, 32, 41, 42, 45, 49, Combellas, Ricardo, 27, 32, 33, 34, 90,
56, 58, 67, 76, 78, 80 137
195
Index
Compensation Act of 1816, 52 Frente Sandinista (FSLN), 28, 136

competing principals, 3, 15, 17, 18, 126, Fujimori, Alberto, 38, 63, 82, 86
127, 132–134, 143, 156, 158, 162,
164, 169 gender quotas, 10, 12
Confederate Congress, 6 Germany, 32, 33, 89
confidence vote, 134, 139–146, 147, 150, Gioja, José Luis, 50
153, 154, 159, 163, 169 governors, 4, 15, 18, 74, 142
Costa Rica, 25, 26, 28, 32, 41, 45, 49, 56, gridlock. See deadlock
57, 59, 60, 61, 64, 71, 78, 79, 136, Guatemala, 31, 45, 56, 57, 58, 105, 107,
137 108, 141
Cox, Gary, 5, 6, 7, 9, 12, 29, 45, 71 hand raising. See signal voting
113, 126, 127, 128, 139, 141 Hix, Simon, 18, 125, 133
172, 175 Hurtado, Carlos, 27, 86, 136, 149
cross-voting. See discipline hybrid regimes, 45, 134, 135, 139, 143,
cycling problem, 5 146, 158
Czech Republic, 99, 102, 103, 104, 105,
108, 140, 141 indiscipline. See discipline
individualism, 2, 12, 14–22, 27, 29, 36,
de la Rua, Fernando, 50 40, 41, 42, 61, 62, 71, 133, 143,
deadlock, 1 157, 166–173, 175
decisiveness, 1, 4–8, 14 instability problem. See cycling problem
Desposato, Scott, 25, 59, 72, 123, 142, interest groups, 4, 15, 39, 54, 73, 74, 80,
163, 172 167
direct democracy, 30 intraparty competition. See
discipline, 1, 7, 8, 24–31, 36, 38, 41, 44, individualism
51, 54, 61, 75, 76, 80, 88, 89, 97, Iraq, 9, 10, 11, 12, 13, 14
126–136, 137, 155, 163, 171 Israel, 103, 104, 106, 107, 108, 109, 110,
disproportionality, 10 111, 140, 141, 142
district magnitude, 10, 12, 71, 72, 98, Italy, 32, 171
123, 133, 168, 173, 175
Japan, 32
Ecuador, 25, 26, 27, 31, 45, 49, 56, 57,
Jefferson, Thomas, 52, 53
58, 67, 71, 78, 105, 107, 108, 141
Jenkins, Jeffrey, 6, 53, 54, 75
El Salvador, 25, 28, 45, 49, 59, 67, 71,
Jordan, 11
75, 77, 78
electronic voting, 46, 49, 51, 54–67 Kwasniewski, Alexander (President of
73, 76–80, 82, 83, 86, 89, 90 Poland), 143
120, 167
European Union, 18 Latin America, 7, 20, 21, 24, 25, 29
30, 31, 32, 33, 36, 40, 41, 44, 46
Farm Bureau, 54 49, 50, 51, 55, 56, 61, 64, 65, 66
federalism, 4, 15, 125, 127, 134, 148, 71, 74, 75, 87, 88, 90, 98, 166
154 167
Federalist Party, 52 Latinobarómetro, 30, 187
floor votes. See legislative votes legislative votes, 3, 33, 42, 48, 49, 66, 68,
France, 103, 105, 107, 108, 139, 141, 69, 70, 73, 74, 76, 87, 87, 92, 103,
142 122, 127, 128, 137, 167, 176
196
Index
manipulation of, 76, 77 party unity, 2, 3, 7, 16–22, 23, 24, 42,

thresholds, 56, 57, 64, 97, 100, 101, 47, 59, 80, 89, 107, 92–116, 120,
106, 113, 115, 120, 122, 143 122, 123, 125–135, 137–147, 151,
Linz, Juan, 31, 134 158–163, 165, 166, 168, 169, 170,
Lula (Luiz Ignacio da Silva, president 175, 176
of Brazil), 155 measurement, 92, 113, 115, 116, 117,
126, 128–131, 132
Mainwaring, Scott, 24, 113, 125, 139, mobilization, 85, 98, 100, 110–116,
142, 162, 163 146, 155, 169, 173
Mayhew, David, 52, 54 partyarchy, 33, 90, 176
Mexico, 25, 28, 31, 41, 45, 56, 58, 59, Pease, Henry, 85
60, 61, 64, 65, 75, 81, 106, 108, Peronist Party (Argentina), 50, 98
137, 141 personal preference votes, 171, 173
mixedmember electoral systems, 13, 33, 41 Peru, 25, 37, 38, 45, 49, 51, 56, 58
Monroe, Burt, 9 60, 61, 63, 64, 71, 82, 85, 86, 88
Morgenstern, Scott, 59, 126 89, 90, 91, 98, 101, 104, 106
multimember district (MMD) 107, 108, 141, 143, 144
elections, 44, 71 Philippines, 32, 106, 107, 108, 131, 141
National Rifle Association, 54 Poland, 103, 106, 107, 108, 141, 142,
New Zealand, 32, 98, 100, 103, 106, 143, 144, 146
107, 108, 141 political parties, 1, 3, 4–7, 13, 20, 24
NGOs, 72, 81, 90 30, 32, 44, 92, 169
Nicaragua, 25, 28, 45, 49, 57, 58, 60, 61, age of, 140
64, 71, 78, 80, 86, 87, 89, 97, 106, dissident legislators, 76
107, 108, 120, 121, 122, 136, 137, dissident members, 21, 53, 73, 74, 75,
141, 143 80, 90
nonvotes, 94–100, 107–122, 146, 154, governing parties, 21, 30, 76, 90, 98,
165, 173 110, 135, 138, 147–158, 161, 169,
nonvoting equilibria, 99 174, 175
leaders, 7, 8, 14–20, 31, 36, 39, 41, 47,
Odegard, Peter, 54 80, 133, 135, 137, 138, 168
Orduz, Rafael, 79, 83 measuring unity in small parties, 85,
101, 102, 122, 154
Panama, 31, 45, 49, 57, 58, 59, 60, 61, party groups, 25, 26, 75, 104, 124,
63, 64, 76, 78, 81, 90 134, 136
Paniagua, Valentı́n, 86 party labels, 1, 7, 33, 93, 140, 170
parliamentarism, 21, 22, 29, 30, 31, 125, partyarchy, 34
135, 138, 139, 142, 149, 153–158, rank-and-file legislators, 46, 53, 54,
161, 162, 169, 171, 174 61, 73, 74, 75, 97, 100, 101, 132, 137
party labels, 8, 17, 30, 32, 169 preference votes, 10, 17, 22, 41, 71, 148,
party leaders, 4, 7, 8, 15, 18, 20–22, 25, 162, 173, 175, see individualism
27, 29, 31, 32, 36–41, 46, 49, presidentialism, 11, 18, 21, 29, 30
53–56, 58, 61, 62, 65, 74, 72–76, 45, 46, 56, 59, 81, 104, 125
79–83, 88, 89, 97, 101, 126–135, 126, 127, 134–139, 143, 146, 147,
137, 149, 157, 162–164, 167 152–159, 161–164, 167, 169, 171,
169, 170 173, 174
197
Index
presidents, 4, 15–18, 24, 39, 40, 74, 82, Samper, Jorge, 28, 86, 87, 136, 137
125, 134, 135, 137, 139, 149, 154, Sanchez de Lozada, Gonzalo, 28, 76, 81
157, 158, 163, 164, 169, 170, 171, secret ballot, 53, 68
175 secret voting, 50, 55
principals, 3, 4, 14–22, 23, 36, 37, 38, separation of powers. See
39, 69, 70, 72, 74, 90, 127, 133, presidentialism
134, 135, 137, 138, 139, 162 signal voting, 50, 70, 72, 74, 77
166–170 single non-transferable vote (SNTV)
proportional representation, 8, 9, 31, elections, 11, 12, 172
143, 168, 172 single-member district (SMD)
closed lists, 8, 10, 41, 125, 133 elections, 9, 13, 44, 71
Proyecto de Elites Latinoamericanas, Sophocles, 20
36 Stokes, Susan, 43, 125, 140, 170
public votes, 44, 45, 46, 49, 50, 51
53, 54, 58, 66, 68, 69, 75, 79, 81 Taiwan, 11
82, 91 Tarek Saab, William, 32, 34, 90
Texas, state legislature, 76, 85, 90
Quinn, Kevin, 46, 47 Toledo, Alejandro (president of Peru),
86, 98
recall elections, 33, 34 transparency, 2, 21, 44, 45, 46, 63, 68,
recorded votes, 21, 22, 36, 46, 47, 51, 69, 74, 78, 80–84, 87, 90, 102, 165,
55, 57, 59, 60, 62, 65, 76, 78, 79 166, 167, 176
80, 81–90, 94, 102, 103, 104 monitoring, 2, 21, 48, 49, 50, 52, 53,
143, 167 54, 59, 70–75, 81, 88, 165, 167
reform, 8, 31 monitoring votes, 21, 50, 54, 55, 56,
regimes, age of, 141, 148 68, 69, 72–75, 83, 167
Republican Party (United States), 24, Web sites, 50, 59, 63, 64, 65, 76, 81
52 Transparency International, 81, 87, 88
responsible party government, 23, 24, U.S. Congress, 6, 7, 47, 49, 51, 52, 53,
91 55, 57, 60, 64, 103, 111, 113, 127,
responsiveness, 1, 3, 8, 29–33, 41, 52, 176
69, 74, 89, 127, 133, 134, 162, 169, Legislative Reorganization Act of
170, 171 1970, 55
rhetorical ideal point estimation, Speaker of the House, 54
46, 47
RICE index, 94, 95, 96, 102, 112, 115, Ukraine, 32
116, 117, 122, 123, 124, 129, 130 ULOSER index, 96, 99, 100, 101, 102,
Rice, Stuart, 94 108, 109, 110, 111, 112, 118, 119,
RLOSER index, 96, 99, 100, 101, 101, 173
102, 108, 109, 110, 111, 112, 118, United States, 21, 24, 43, 44, 45, 49
119, 120, 154 51, 52, 54, 55, 57, 58, 65, 68, 74
roll call voting. See recorded votes 76, 85, 86, 87, 91, 103, 104, 106,
Rose-Ackerman, Susan, 43, 171 107, 108, 110, 112, 116, 141
Rousseau, Jean Jacques, 68, 69, 91 144, 167, 176
Russia, 32, 97, 103, 104, 106, 107, 108, UNITY index, 95, 96, 102, 112, 115,
120, 141, 143 116, 117, 123
198
Index
Uribe, Alvaro (presidnt of Colombia), 1999 Constitution, 27, 33

82 visible votes, 21, 45, 51, 55, 61, 62, 72,
Uruguay, 45, 56, 57, 58, 59, 104, 106, 73, 74, 79, 84, 87
107, 108, 141, 144 vote records, 2, 44, 56, 59, 60, 61, 104
VoteWorld, 103
Vanuatu, 11
Venezuela, 25, 27, 31, 32, 33, 34, 37, 38, Weldon, Jeffrey, 31, 41, 127, 137
41, 42, 45, 49, 56, 58, 60, 61, 64, 78, WRICE index, 99, 100, 101, 102, 108,
90, 137 109, 110, 111, 112, 121, 130, 160
199
Other Books in the Series (continued from page iii)
Catherine Boone, Political Topographies of the African State: Territorial

Authority and Institutional Change
Michael Bratton and Nicolas van de Walle, Democratic Experiments in
Africa: Regime Transitions in Comparative Perspective
Michael Bratton, Robert Mattes, and E. Gyimah-Boadi, Public Opinion,
Democracy, and Market Reform in Africa
Valerie Bunce, Leaving Socialism and Leaving the State: The End of Yugoslavia,
the Soviet Union, and Czechoslovakia
Daniela Caramani, The Nationalization of Politics: The Formation of
National Electorates and Party Systems in Europe
Kanchan Chandra, Why Ethnic Parties Succeed: Patronage and Ethnic Head-
counts in India
José Antonio Cheibub, Presidentialism, Parliamentarism, and Democracy
Ruth Berins Collier, Paths toward Democracy: The Working Class and Elites in
Western Europe and South America
Christian Davenport, State Repression and the Domestic Democratic Peace
Donatella della Porta, Social Movements, Political Violence, and the State
Alberto Diaz-Cayeros, Federalism, Fiscal Authority, and Centralization in
Latin America
Gerald Easter, Reconstructing the State: Personal Networks and Elite Identity
M. Steven Fish, Democracy Derailed in Russia: The Failure of Open Politics
Robert F. Franzese, Macroeconomic Policies of Developed Democracies
Roberto Franzosi, The Puzzle of Strikes: Class and State Strategies in Postwar
Italy
Geoffrey Garrett, Partisan Politics in the Global Economy
Miriam Golden, Heroic Defeats: The Politics of Job Loss
Jeff Goodwin, No Other Way Out: States and Revolutionary Movements
Merilee Serrill Grindle, Changing the State
Anna Grzymala-Busse, Rebuilding Leviathan: Party Competition and State
Exploitation in Post-Communist Democracies
Anna Grzymala-Busse, Redeeming the Communist Past: The Regeneration of
Communist Parties in East Central Europe
Frances Hagopian, Traditional Politics and Regime Change in Brazil
Gretchen Helmke, Courts under Constraints: Judges, Generals, and Presidents
in Argentina
Yoshiko Herrera, Imagined Economies: The Sources of Russian Regionalism
J. Rogers Hollingsworth and Robert Boyer, eds., Contemporary Capitalism:
The Embeddedness of Institutions
John D. Huber and Charles R. Shipan, Deliberate Discretion? The Institutional
Foundations of Bureaucratic Autonomy
Ellen Immergut, Health Politics: Interests and Institutions in Western Europe
Torben Iversen, Capitalism, Democracy, and Welfare
Torben Iversen, Contested Economic Institutions
Torben Iversen, Jonas Pontussen, and David Soskice, eds., Unions,
Employers, and Central Banks: Macroeconomic Coordination and Institu-
tional Change in Social Market Economies
Thomas Janoski and Alexander M. Hicks, eds., The Comparative Political
Economy of the Welfare State
Joseph Jupille, Procedural Politics: Issues, Influence, and Institutional Choice in
the European Union
Stathis Kalyvas, The Logic of Violence in Civil War
David C. Kang, Crony Capitalism: Corruption and Capitalism in South Korea
and Philippines
Junko Kato, Regressive Taxation and the Welfare State
Robert O. Keohane and Helen B. Milner, eds., Internationalization and
Domestic Politics
Herbert Kitschelt, The Transformation of European Social Democracy
Herbert Kitschelt, Peter Lange, Gary Marks, and John D. Stephens, eds.,
Continuity and Change in Contemporary Capitalism
Herbert Kitschelt, Zdenka Mansfeldova, Radek Markowski, and Gabor
Toka, Post-Communist Party Systems
David Knoke, Franz Urban Pappi, Jeffrey Broadbent, and Yutaka
Tsujinaka, eds., Comparing Policy Networks
Allan Kornberg and Harold D. Clarke, Citizens and Community: Political
Support in a Representative Democracy
Amie Kreppel, The European Parliament and the Supranational Party System
David D. Laitin, Language Repertories and State Construction in Africa
Fabrice E. Lehoucq and Ivan Molina, Stuffing the Ballot Box: Fraud, Electoral
Reform, and Democratization in Costa Rica
Mark Irving Lichbach and Alan S. Zuckerman, eds., Comparative Politics:
Rationality, Culture, and Structure, Second Edition
Evan Lieberman, Race and Regionalism in the Politics of Taxation in Brazil and
South Africa
Julia Lynch, Age in the Welfare State: The Origins of Social Spending on
Pensioners, Workers, and Children
Pauline Jones Luong, Institutional Change and Political Continuity in Post-Soviet
Central Asia
Doug McAdam, John McCarthy, and Mayer Zald, eds., Comparative
Perspectives on Social Movements
Beatriz Magaloni, Voting for Autocracy: Hegemonic Party Survival and Its
Demise in Mexico
James Mahoney and Dietrich Rueschemeyer, eds., Historical Analysis and the
Social Sciences
Scott Mainwaring and Matthew Soberg Shugart, eds., Presidentialism and
Democracy in Latin America
Isabela Mares, The Politics of Social Risk: Business and Welfare State Development
Isabela Mares, Taxation, Wage Bargaining, and Unemployment
Anthony W. Marx, Making Race, Making Nations: A Comparison of South
Africa, the United States, and Brazil
Bonnie Meguid, Competition between Unequals: The Role of Mainstream
Parties in Late-Century Africa
Joel S. Migdal, State in Society: Studying How States and Societies Constitute
One Another
Joel S. Migdal, Atul Kohli, and Vivienne Shue, eds., State Power and Social
Forces: Domination and Transformation in the Third World
Scott Morgenstern and Benito Nacif, eds., Legislative Politics in Latin America
Layna Mosley, Global Capital and National Governments
Wolfgang C. Muller and Kaare Strøm, Policy, Office, or Votes
Maria Victoria Murillo, Labor Unions, Partisan Coalitions, and Market
Reforms in Latin America
Ton Notermans, Money, Markets, and the State: Social Democratic Economic
Policies since 1918
Anı́bal Pérez-Liñán, Presidential Impeachment and New Political Instability in
Latin America
Roger Petersen, Understanding Ethnic Violence: Fear, Hatred, and Resentment
in Twentieth-Century Eastern Europe
Simona Piattoni, ed., Clientelism, Interests, and Democratic Representation
Paul Pierson, Dismantling the Welfare State? Reagan, Thatcher, and the
Politics of Retrenchment
Marino Regini, Uncertain Boundaries: The Social and Political Construction of
European Economies
Marc Howard Ross, Cultural Contestation in Ethnic Conflict
Lyle Scruggs, Sustaining Abundance: Environmental Performance in Industrial
Democracies
Jefferey M. Sellers, Governing from Below: Urban Regions and the Global
Economy
Yossi Shain and Juan Linz, eds., Interim Governments and Democratic
Transitions
Beverly Silver, Forces of Labor: Workers’ Movements and Globalization since
1870
Theda Skocpol, Social Revolutions in the Modern World
Regina Smyth, Candidate Strategies and Electoral Competition in the Russian
Federation: Democracy without Foundation
Richard Snyder, Politics after Neoliberalism: Reregulation in Mexico
David Stark and László Bruszt, Postsocialist Pathways: Transforming
Politics and Property in East Central Europe
Sven Steinmo, Kathleen Thelen, and Frank Longstreth, eds., Structuring
Politics: Historical Institutionalism in Comparative Analysis
Susan C. Stokes, ed., Public Support for Market Reforms in New Democracies
Duane Swank, Global Capital, Political Institutions, and Policy Change in
Developed Welfare States
Sidney Tarrow, Power in Movement: Social Movements and Contentious Politics
Kathleen Thelen, How Institutions Evolve: The Political Economy of Skills in
Germany, Britain, the United States, and Japan
Charles Tilly, Trust and Rule
Daniel Treisman, The Architecture of Government: Rethinking Political
Decentralization
Lily Lee Tsai, Accountability without Democracy: How Solidary Groups Provide
Public Goods in Rural China
Joshua Tucker, Regional Economic Voting: Russia, Poland, Hungary, Slovakia,
and the Czech Republic, 1990–1999
Ashutosh Varshney, Democracy, Development, and the Countryside
Jeremy M. Weinstein, Inside Rebellion: The Politics of Insurgent Violence
Stephen I. Wilkinson, Votes and Violence: Electoral Competition and Ethnic
Riots in India
Jason Wittenberg, Crucibles of Political Loyalty: Church Institutions and
Electoral Continuity in Hungary
Elisabeth J. Wood, Forging Democracy from Below: Insurgent Transitions in
South Africa and El Salvador
Elisabeth J. Wood, Insurgent Collective Action and Civil War in El Salvador

(Cambridge Studies in Comparative Politics) John M. Carey-Legislative Voting and Accountability-Cambridge University Press (2008)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Cambridge Studies in Comparative Politics) John M. Carey-Legislative Voting and Accountability-Cambridge University Press (2008)

Uploaded by

Copyright:

Available Formats

This page intentionally left blank

Legislative Voting and Accountability

Legislatures are the core representative institutions in modern democracies.

John M. Carey is the John Wentworth Professor in the Social Sciences at

Assistant General Editor

Other Books in the Series

Continued after the Index

Cambridge University Press

This publication is in copyright. Subject to statutory exception and to the

ISBN-13 978-0-511-48077-5 eBook (NetLibrary)

ISBN-13 978-0-521-88493-8 hardback

ISBN-13 978-0-521-71191-3 paperback

Cambridge University Press has no responsibility for the persistence or accuracy

1 TO WHOM ARE LEGISLATORS

4 DEMAND FOR VISIBLE VOTES 68

Appendix: Interview Subjects by Country 177

Desposato, Argelina Figuereido, Wendy Hunter, Eduardo Leoni, Fernando

Beyond these, a number of scholars provided comments and suggestions

To Whom Are Legislators Accountable?

legislator’s actions be available to those outside the legislature and capable

principals’ preferences and minimize corruption and other abuses of power

1.1.2.2. Principals. Principals are political actors who command some

1.1.2.3. Decisiveness. Legislative decisiveness refers to the capacity of

1.2. Decisiveness Problems

1.2.1. Bottlenecks and Cycling

present a specific challenge. The number of policy options available in any

1.2.2. Parties and Legislative Action

be mapped in terms of party voting unity.1 If parties are highly unified in

1.3. Collective versus Individual Accountability

1.3.1. Competing Visions

when the demands of leaders conflict with responsiveness to local constit-

1.3.2. Excursus on Electoral Rules: Iraq versus Afghanistan

1.3.2.1. Iraq. Iraq elected two legislatures in 2005 – first, in January,

October plebiscite; then a National Assembly to serve a full four-year term

average (Inter-Parliamentary Union 2006). In sum, by the benchmarks

1.3.2.2. Afghanistan. The Afghan experience with establishing

and gender. Afghan rules lean toward privileging connections between

1.3.3. Collectivism versus Individualism, Not Proportional

(Shugart and Wattenberg 2001; International Institute for Democracy and

1.4. Legislators, Principals, and the Structure of Accountability

1.4.1. Party Leaders and Everyone Else

In short, because of the ubiquity of parties, even the individualistic

1.4.2. Competing Principals and Individualism

and the ideas of collective-versus-individualistic representation. Given the

connection to voters, presidential influence adds another, potentially com-

PARTY PRESIDENT GOVERNOR

European Union Parliament

NATIONAL PARTY EU PARTY

1.5. Plan of the Book

Chapters 3 and 4 explore the conditions under which accountability

Collective Accountability and Its Discontents

2.1. The Strong-Party Ideal

Current observers might argue that, in the closing decades of the

Wrapping up a broad survey of the state of political parties in Latin

2.2. Legislative Parties and Discipline in Latin America

2.2.1. How Parties’ Positions Are Determined

(Acosta, Devia, Garcı́a, Gómez Gallo, Navarro, and Holguı́n interviews).

In explaining the sources of intrapartisan divisions, interview subjects

2.2.3. How and When Sanctions Are Imposed

noted that expulsion is used only if a breach of discipline is pivotal to the

Academic accounts, as well as those of other interview subjects (Vargas

Baltodano correctly anticipated continued conflicts with party leaders

2.3. Trouble in Paradise: Partisan Representation Falling Short

2.3.1. The Costs of Collective Reputations